BREAKING: rename the public apparent-wind API to va / va_vec / va_dist

[1-Bort-1]

BREAKING: rename the public apparent-wind API to va / va_vec / va_dist

Second of the two PRs Bart chose with option A on #337 (comment 5705262071). Stack it on the PR from branch agent/337-one-name-per-apparent-wind-quantity-va-v, which renamed every local, argument, test, example and doc and left the public names alone.

Convention (#147): va = speed [m/s], va_vec = 3-vector, va_dist = per-panel speeds, va_vec_dist = per-panel P×3 vectors.

To rename, with a BREAKING changelog entry:

• body_aero.va property (getproperty/setproperty!) -> va_vec; field _va -> va_vec (drops the _ prefix) and has_distributed_va
• va= keyword of BodyAerodynamics(...) and reinit! -> va_vec=
• Panel.va field -> va_vec
• VSMSolution._va_dist -> va_vec_dist; va_unrefined_dist -> va_vec_unrefined_dist
• BaseResult.va_norm_dist -> va_dist
• v_a_dist fields on BodyAerodynamics and LoopResult: this is |v_rel x y_airf|, not the apparent wind speed; locals are already v_rel_dist
• SemiInfiniteFilament.vel_mag -> va
• v_a= keyword of plot_polars / plot_combined_analysis -> va=
• linearize va_idxs= -> va_vec_idxs=, and the "va_ref" result key, which holds a vector
• YAML airfoil.v_app (AirfoilSettings.v_app) -> va, and condition.wind_speed (ConditionSettings), which is also the apparent wind speed

Before writing code, ask whether each renamed name should keep a depwarn alias. Option C (deprecations) was not chosen, but body_aero.va silently switching to a scalar would be worse than an error. SymbolicAWEModels reads ._va, panel.va and solver.lr.v_a_dist, so it needs a follow-up PR.

[1-Bort-1]

Renaming the public apparent-wind names listed here (body_aero.va/_va, va= keywords, Panel.va, VSMSolution._va_dist, va_norm_dist, v_a_dist, vel_mag, plot v_a=, linearize va_idxs/va_ref, YAML v_app/wind_speed) with a BREAKING changelog entry, stacked on #349. First I will ask which old names keep a depwarn alias. SymbolicAWEModels' follow-up is out of scope here.

[1-Bort-1]

Question from VortexStepMethod.jl-348

Should any of the renamed public names keep a depwarn alias for one release?

None of them can silently turn into a scalar: this PR adds no scalar body_aero.va, no scalar va field on Panel and no va= keyword on BodyAerodynamics or reinit!. Old code fails loudly instead:

• body_aero.va and panel.va give a FieldError.
• va= on BodyAerodynamics/reinit!, v_a= on plot_polars/plot_combined_analysis, and va_idxs= on linearize give an 'unsupported keyword argument' MethodError.
• The old YAML keys wind_speed and v_app give 'FieldError: type ConditionSettings has no field wind_speed, available fields: va, ...', from StructMapping as it stands. The repo's own 7 settings YAMLs get renamed here.
• result["va_ref"] gives a KeyError. Nothing in src, test, examples or ext reads that key.
  The one catch is plot_polars(...; va=...): it keeps a keyword named va that is a speed, but it was a speed before as well (v_a=).

A) No aliases. Every old name errors, BREAKING changelog entry, and SymbolicAWEModels gets its follow-up PR.
B) A depwarn alias only for body_aero.va and the va= keyword on BodyAerodynamics/reinit!, the vector names downstream code uses most. That means a getproperty branch and a keyword check to remove in v7.
C) Depwarns for every renamed name, including the YAML keys. That is option C from #337 again, about 60 lines of shims.

Unless you say otherwise, I will also use v_a_dist -> v_rel_dist, matching the locals and the calculate_results docstring, and "va_ref" -> "va_ref_vec", matching its local.

My recommendation: A. Nothing silently changes meaning, and the errors already name the new field.