Color pipeline P10: hardware measurement gate and program closure

Phase P10 of the color pipeline implementation (see parent issue). Program-level closure gate: per #4032, static datasheet analysis alone cannot close the implementation phase — measured hardware evidence is required.

Scope

Measured ground truth, compatibility hardening, and program sign-off.

• Instrument measurement (A2, B9): at least one representative WS2812B/SK6812-class strip and one SPI/high-bit-depth strip, across primaries, secondaries, neutral ramp, low codes, full scale, operating temperature. Primary/secondary chromaticity via spectroradiometer (≤ ±0.002 xy on narrowband) or spectrally corrected colorimeter; plain colorimeter only for neutral/luminance tracking. Fixture: fixed geometry, dark surround, logged temperature and drive current.
• Measurement reports + measured profiles (C7): reports (instrument model/serial, date, fixture, temperature, raw data) checked into FastLED/datasheets, append-only; measured profiles minted with new report IDs (C8.2) and ingested via the P3 pipeline as a reviewed PR. Profile-accuracy budget: median ΔE2000 ≤ 1.5 / max ≤ 4.0 at reference temperature (A1); thermal drift recorded as documentation (B10). Re-measurement triggers documented: new part, bin divergence beyond budget, out-of-budget validation.
• Legacy compatibility: byte-identical output with color management disabled, proven across representative chipsets and the existing test corpus; the B6 documentation (legacy .fled appearance changes only with CM enabled, and why) shipped in user-facing docs.
• Program sign-off: full C++ test suite green; representative hardware validation via examples/AutoResearch/AutoResearch.ino + bash autoresearch (no new example sketches); every parent acceptance-criteria row checked or explicitly re-assigned by a follow-up decision issue.

Decisions implemented

A1 (profile budgets), A2, B9, B10, C7; closes the loop on C8/C9 by exercising the mint→ingest→bind path with real measured data.

Acceptance criteria owned

• Instrument measurement of WS2812B/SK6812-class + one high-bit-depth strip.
• Legacy output byte-identical with CM disabled.
• Full C++ tests + representative hardware validation green.

Dependencies

All prior phases (P1–P9).

Deliverables

Measurement reports + measured profiles in FastLED/datasheets, ingested generated profiles, byte-identical regression suite, hardware validation evidence, final program report against the parent's criteria table.