fix: name norm_coords() arguments at the rglplot.igraph() call site

[krlmlr]

Fixes the sole red leg of the scheduled full rcc matrix on main,
rcc: ubuntu-26.04 (4.6) with lifecycle errors:

── Error ('test-plot.R:85:3'): rglplot() works ──
Error: Calling `norm_coords()` with positional or abbreviated arguments was deprecated in igraph 3.0.0.
 11.   └─igraph::norm_coords(layout, -1, 1, -1, 1, -1, 1) at igraph/R/plot.R:1626:5

#2799 named the arguments at plot.igraph()'s 2-D call site but missed
the 3-D one in rglplot.igraph().
The leg only runs on schedule or manual dispatch
(the versions-matrix step is skipped on same-repo PRs),
so PR runs never surfaced it.
The fix names xmin/xmax/ymin/ymax/zmin/zmax at the caller —
no verbosity pin, per the fix-the-caller policy.

Validation (with libglu1-mesa + rgl installed so rglplot() works
actually runs instead of skipping):
full suite FAIL 0 both plain and with IGRAPH_LIFECYCLE_ERRORS=true;
generator, air, and devtools::document() are no-ops.

Not addressed here (deliberately):
the nightly Sanitizer failure
(R_getRegisteredNamespace undeclared on the CSAN image)
was removed from this PR per maintainer request
and remains open;
the Bioconductor graph pak outage and the sandbox-only
graphsdb fetch are environmental.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 4cb4a64 is merged into main:

• ✔️as_adjacency_matrix: 771ms -> 769ms [-0.79%, +0.3%]
• 🚀as_biadjacency_matrix: 751ms -> 746ms [-1.16%, -0.01%]
• ✔️as_data_frame_both: 1.52ms -> 1.55ms [-0.35%, +4.14%]
• ✔️as_long_data_frame: 4.02ms -> 4.05ms [-1.81%, +3.22%]
• ✔️es_attr_filter: 2.73ms -> 2.7ms [-3.01%, +0.77%]
• ✔️graph_from_adjacency_matrix: 121ms -> 120ms [-1.85%, +0.83%]
• ❗🐌graph_from_data_frame: 3.42ms -> 3.49ms [+0.37%, +3.95%]
• ✔️vs_attr_filter: 1.53ms -> 1.54ms [-0.82%, +2.95%]
• ✔️vs_by_name: 982µs -> 998µs [-2.11%, +5.26%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if d3bdec3 is merged into main:

• ✔️as_adjacency_matrix: 886ms -> 876ms [-3.47%, +1.29%]
• ✔️as_biadjacency_matrix: 874ms -> 889ms [-0.48%, +3.71%]
• ✔️as_data_frame_both: 1.68ms -> 1.71ms [-3.88%, +8.02%]
• ✔️as_long_data_frame: 4.47ms -> 4.67ms [-0.35%, +8.96%]
• ✔️es_attr_filter: 3.15ms -> 3.12ms [-4.48%, +2.69%]
• ✔️graph_from_adjacency_matrix: 157ms -> 160ms [-3.15%, +6.85%]
• ✔️graph_from_data_frame: 3.96ms -> 3.94ms [-3%, +1.77%]
• ✔️vs_attr_filter: 1.87ms -> 1.86ms [-3.49%, +2.06%]
• ✔️vs_by_name: 1.21ms -> 1.18ms [-6.41%, +1.31%]
  Further explanation regarding interpretation and methodology can be found in the documentation.