feat: expose XFOIL Type 2/3 polar modes (variable Re)

[aeronauty-flexcompute]

Summary

• Adds ReType enum (Type 1/2/3) to the Rust solver with effective_reynolds() that rescales Re each viscous iteration based on CL — matching original XFOIL Fortran behaviour
• Threads re_type parameter through Python bindings (analyze_faithful, analyze_faithful_batch, get_bl_distribution) and the flexfoil API (solve(), polar(), bl_distribution())
• Returns reynolds_eff in all viscous results so callers can see the effective Re used
• Type 1 (constant Re) remains the default everywhere — no breaking changes

Requested by Ferdinand Elsner (KIT Karlsruhe) for GA-based airfoil optimisation where the operating Re depends on CL.

Usage

import flexfoil

foil = flexfoil.naca("2412")

# Type 1 (default, unchanged)
r1 = foil.solve(5.0, Re=1e6)

# Type 2: fixed Re*sqrt(CL)
r2 = foil.solve(5.0, Re=1e6, re_type=2)
print(r2.reynolds_eff)  # effective Re at converged CL

# Type 3: fixed Re*CL
r3 = foil.solve(5.0, Re=1e6, re_type=3)

# Works with polar sweeps too
polar = foil.polar(alpha=[0, 3, 5, 7, 10], Re=1e6, re_type=2)

Test plan

• 6 Rust unit tests for effective_reynolds() (Type 1/2/3, negative CL, near-zero CL guard)
• 10 Python binding-level tests (test_solver.py::TestReType)
• 8 Python API-level tests (test_api.py::TestReType)
• All 105 existing tests pass — no regressions
• Docstrings and docs-site API reference not yet updated (planned for follow-up)

🤖 Generated with Claude Code