docs: label the performance tables 4.0.0 and fix two stale sentences

[lesnik512]

Summary

Prepares docs/introduction/performance.md for the 4.0.0 tag.

• The Results paragraph and the "what moved" table column now say 4.0.0 instead of "the end of the Refactor and perf: registry hooks, container-owned cache and context, cheaper child build #605 perf series (the 4.0 code, unreleased when measured)" and "4.0 with Refactor and perf: registry hooks, container-owned cache and context, cheaper child build #605". The tables were measured on the Refactor and perf: registry hooks, container-owned cache and context, cheaper child build #605 series; the code on main has not changed since, so they describe 4.0.0 as long as nothing else lands before the tag.
• The by-type vs by-reference sentence quoted old cells (146 vs 142 on C2, 760 vs 764 on C3, "a few nanoseconds"). It now quotes the published cells: 242 vs 239, 143 vs 140, 756 vs 772, a 3-16 ns spread.
• The C4 paragraph said levelling the by-type asymmetry "would not move the dishka ratio (1.12) below 1.0". That ratio is now 0.88. The paragraph now states the size of the asymmetry (at most 16 ns, under 1% of the 1.85 µs C4 cell) and that it cannot flip any C4 ratio across 1.0. It also drops the claim that the asymmetry cuts against modern-di, since the by-type cell is now faster on C1 and C2 and slower on C3.

Test plan

• just lint-ci
• tests/test_bench_report.py, tests/test_docs_slug_census.py
• mkdocs build --strict

[github-actions]

Benchmark

Details

Benchmark suite	Current: 8aaa608	Previous: 9131e81	Ratio
benchmarks/test_guard_by_type.py::test_g16_resolve_by_type	2888590.6964573544 iter/sec (stddev: 1.1659785543007837e-8)	2890422.8555649277 iter/sec (stddev: 2.171322569513238e-8)	1.00
benchmarks/test_guard_by_type.py::test_g17_resolve_by_type_large_registry	2747265.4269382046 iter/sec (stddev: 2.4742183873926905e-8)	2824226.8163719093 iter/sec (stddev: 1.9328131046109683e-8)	1.03
benchmarks/test_guard_cold.py::test_g8_cold_first_resolve	21595.11096205255 iter/sec (stddev: 0.00004183906613727969)	21794.7634610908 iter/sec (stddev: 0.0000418033427966709)	1.01
benchmarks/test_guard_cold.py::test_g8b_cold_first_resolve_cached	16519.674759705726 iter/sec (stddev: 0.00010744990777293124)	16435.91819346946 iter/sec (stddev: 0.00012635225158853428)	0.99
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[1]	364.18731038951086 iter/sec (stddev: 0.00003754526479311828)	446.4820096558723 iter/sec (stddev: 0.00005765809189429768)	1.23
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[2]	347.66885979028945 iter/sec (stddev: 0.00009901342633923239)	408.8667276632547 iter/sec (stddev: 0.00010451882649088393)	1.18
benchmarks/test_guard_concurrency.py::test_g14_concurrent_cached_hit[4]	337.1792390886368 iter/sec (stddev: 0.00006830677524244392)	396.45852955547485 iter/sec (stddev: 0.0000854385841269237)	1.18
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[1]	13024.964406573637 iter/sec (stddev: 0.000008704837049042389)	11551.86948232998 iter/sec (stddev: 0.000012245905907817008)	0.89
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[2]	6204.88980475804 iter/sec (stddev: 0.000013733025807173439)	5842.371928992844 iter/sec (stddev: 0.00002290193155139606)	0.94
benchmarks/test_guard_concurrency.py::test_g15_concurrent_first_resolve[4]	3384.1179453092773 iter/sec (stddev: 0.000039810431144514364)	2880.8207343012486 iter/sec (stddev: 0.00012677251323193323)	0.85
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[1]	8468.673318798992 iter/sec (stddev: 0.000010496734128653497)	6732.055495708335 iter/sec (stddev: 0.000033304820243355345)	0.79
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[2]	3687.6197631223104 iter/sec (stddev: 0.000025622813735866334)	3047.853742035851 iter/sec (stddev: 0.00007234744876680933)	0.83
benchmarks/test_guard_concurrency.py::test_g15b_concurrent_first_resolve_sibling_children[4]	1726.5547355633346 iter/sec (stddev: 0.00004538902842327305)	1679.4412196773526 iter/sec (stddev: 0.00008598878115642822)	0.97
benchmarks/test_guard_concurrency.py::test_g15c_worker_pool_floor_control[1]	41356.42170608497 iter/sec (stddev: 0.000004499497775486151)	41865.77622722524 iter/sec (stddev: 0.000004121967760673373)	1.01
benchmarks/test_guard_concurrency.py::test_g15c_worker_pool_floor_control[2]	13129.253673121219 iter/sec (stddev: 0.000010639718763676391)	14871.934057771345 iter/sec (stddev: 0.000017458752167632775)	1.13
benchmarks/test_guard_concurrency.py::test_g15c_worker_pool_floor_control[4]	5856.967866259543 iter/sec (stddev: 0.000030704510733239856)	6001.449350021876 iter/sec (stddev: 0.000028271118570301903)	1.02
benchmarks/test_guard_lifecycle.py::test_g6_build_child_container	1590040.305530693 iter/sec (stddev: 3.187198105965422e-8)	1503713.382610027 iter/sec (stddev: 1.2589515832151697e-7)	0.95
benchmarks/test_guard_lifecycle.py::test_g6b_build_child_container_auto_scope	1675198.5926903165 iter/sec (stddev: 1.8415943666037785e-8)	1603783.3248649265 iter/sec (stddev: 8.426751676796289e-8)	0.96
benchmarks/test_guard_lifecycle.py::test_g7_request_lifecycle_batch	2859.201653565273 iter/sec (stddev: 0.000011038469285263764)	2814.405309585034 iter/sec (stddev: 0.000028038457387894606)	0.98
benchmarks/test_guard_lifecycle.py::test_g7c_event_loop_floor_control	58454.41979625808 iter/sec (stddev: 0.000001864119899855474)	57047.21874111053 iter/sec (stddev: 0.0000025400641199274363)	0.98
benchmarks/test_guard_lifecycle.py::test_g7b_request_cycle_sync	337494.1297110103 iter/sec (stddev: 9.853528297676007e-8)	337700.5101002035 iter/sec (stddev: 8.887268724183936e-8)	1.00
benchmarks/test_guard_lifecycle.py::test_g13_teardown_at_scale	49684.02762603625 iter/sec (stddev: 0.0000017861895905881524)	48883.11114594868 iter/sec (stddev: 0.0000018113985494011498)	0.98
benchmarks/test_guard_lifecycle.py::test_g13b_teardown_at_scale_async_no_finalizers	479.834252101133 iter/sec (stddev: 0.0006709754833864612)	472.851179965394 iter/sec (stddev: 0.0010531926200247062)	0.99
benchmarks/test_guard_resolve.py::test_g1_transient_resolve	1999316.5336361898 iter/sec (stddev: 2.259787336589669e-8)	2035924.9130481875 iter/sec (stddev: 3.116605395268908e-8)	1.02
benchmarks/test_guard_resolve.py::test_g2_cached_resolve	2776853.471236506 iter/sec (stddev: 9.041020417455966e-9)	2825046.135828985 iter/sec (stddev: 2.6765156102711058e-8)	1.02
benchmarks/test_guard_resolve.py::test_g3_deep_chain	692488.6209405486 iter/sec (stddev: 4.206290114449948e-8)	658286.6772658118 iter/sec (stddev: 4.41420259029746e-8)	0.95
benchmarks/test_guard_resolve.py::test_g4_wide_resolve	458702.3612920606 iter/sec (stddev: 3.527033938807588e-7)	444961.87577797804 iter/sec (stddev: 4.2588901059265685e-7)	0.97
benchmarks/test_guard_resolve.py::test_g5_cross_scope	1872698.979653147 iter/sec (stddev: 2.9443693850583948e-8)	1838096.4232320657 iter/sec (stddev: 3.381790367696809e-8)	0.98
benchmarks/test_guard_resolve.py::test_g9_context_resolve	1337119.3421784786 iter/sec (stddev: 1.2146123749682402e-7)	1360563.6080320515 iter/sec (stddev: 1.653562330719312e-7)	1.02
benchmarks/test_guard_resolve.py::test_g12_override_active_resolve	705508.5579052843 iter/sec (stddev: 4.3360607224393286e-8)	695073.8241399634 iter/sec (stddev: 4.159015754803448e-8)	0.99
benchmarks/test_guard_resolve.py::test_g18_alias_hop	2776107.5635133614 iter/sec (stddev: 1.023106160184092e-8)	3219062.9301397656 iter/sec (stddev: 1.8492908297390954e-8)	1.16
benchmarks/test_guard_validate.py::test_g10_validate_deep_chain	27801.24485994631 iter/sec (stddev: 0.00002104406027411718)	27343.970994836855 iter/sec (stddev: 0.000022147211121743528)	0.98
benchmarks/test_guard_validate.py::test_g11_validate_wide	15819.816592014822 iter/sec (stddev: 0.00014882032458866007)	16188.088406352832 iter/sec (stddev: 0.000027908285299662477)	1.02

This comment was automatically generated by workflow using github-action-benchmark.