README: fold P6's BLER mechanism finding into the RLLA/LTSLA headline

Extends the multi-seed gain table with why each arm under/over-performs,
not just the bare percentages: RLLA's realized BLER runs at 0.13x-0.17x
of the 10% target across every config (massive conservatism, likely an
EpsilonGreedyRLLA MCS-selection calibration issue, not a weaker-algorithm
story); LTSLA matches OLLA's BLER at 3 km/h (a genuine efficiency win)
but overshoots to 1.57x-2.07x of target at 30/120 km/h, worsening
monotonically with speed. Flags the live-Doppler forgetting factor as an
open question, not a conclusion. Full table and methodology:
docs/redesign/P6_reaudit_status.md SS8.

Author: karrthikmohank

README.md

@@ -467,6 +467,29 @@ LTSLA's effect is regime-dependent and consistently reproduced across all
 5 seeds: favorable at **3 km/h** (CDL-A +6.68%, CDL-C +10.06%), unfavorable
 at 30/120 km/h (−2.5% to −11.0%). RLLA underperforms OLLA throughout.
 
+**Why**: a BLER diagnostic (same 5 seeds, N_TTIS=30000) traces each arm's
+gain to a distinct mechanism rather than leaving it as a bare percentage.
+OLLA regulates to 0.99×–1.01× of the 10% target at every config, validating
+it as the baseline. **RLLA**'s realized BLER runs at just **0.13×–0.17×**
+of that target across every config (1.3%–1.7% actual) — a massive
+conservatism, not a reasonable-but-suboptimal tradeoff: it isn't failing to
+decode, it is picking far more robust MCS indices than necessary, likely a
+design/calibration issue in `EpsilonGreedyRLLA`'s MCS selection (which
+specific parameter is unconfirmed). **LTSLA** splits by regime: at 3 km/h
+its BLER matches OLLA's (0.90×–1.02× of target) — a genuine
+spectral-efficiency win at matched reliability — but at 30/120 km/h it
+overshoots to **1.57×–2.07×** of target, worsening monotonically with
+speed, meaning the loss there is an under-regulated reliability problem,
+not merely a suboptimal-but-safe MCS choice. Open question, not concluded
+here: the `1236873` fix corrected a stale static-lookup-table bug in the
+Doppler-adaptive forgetting factor, but this data suggests its live-Doppler
+replacement may itself be under-compensating as speed increases — a
+candidate for future investigation, not resolved by this measurement.
+Full per-config BLER table and methodology:
+[`docs/redesign/P6_reaudit_status.md`](docs/redesign/P6_reaudit_status.md) §8
+(this verification covers only P6's own 6-config × 3-arm grid — no other
+project's committed data has an equivalent BLER-vs-target check).
+
 Note: n=5 seeds; no claim of statistical significance beyond this sample
 size.