docs(persona): A1-A5 fresh end-to-end walkthroughs (audit #69 follow-up)

[amaster97]

Summary

• Five player-POV walkthroughs at distinct stack depths and hand types, filling the audit-flagged gap where the only persona-test walkthrough on main was J7o at 40 BB (over-used).
• All five use the 169-class True Path B fast engine (PR feat(engine): true 169-class abstraction mode (Premium-A Phase 1.5 unblock, #68) #171) at 10k DCFR iters; postflop equity via Monte Carlo (poker_solver.equity.equity). Format mirrors the existing J7o walkthrough doc on main.
• Honest reporting: A1 flags engine's preferred 4x open vs spec's 3x and engine's 39 BB 4-bet vs spec's 21 BB; A5 surfaces a finding about middling-pair 3-bet frequency at oversized opens.

Walkthroughs

#	Hand	Stack	Line	Headline
A1	A♠K♦	100 BB	open 3x → 3-bet → 4-bet → call → J83r	Premium 4-bet sequence; flop equity 54% vs call-vs-4bet range
A2	7♠8♠	60 BB	SB limp → BB check → K95r	87s limps 100% at 60 BB (GTO); flop equity 49% (OESD + BD flush)
A3	4♠4♦	80 BB	SB open 3x → BB call → T72r	Set-mining math holds; flop equity 55% (under-pair); 95% (set-hit)
A4	2♦7♥	200 BB	SB 5x → BB defend → Q83r	Crash check PASS; BB defends 48.8%; 27o equity 14.6% (~15% predicted)
A5	BB-defending	80 BB	BB vs 4 BB open	Premium-cell match 15/15 = 100%; avg premium fold 0%

Findings

• All five walkthroughs ran without engine crashes
• Premium hand (AKo, AA) preflop sequences behave as expected
• A1 finding: engine prefers larger open sizes (4x) and larger 4-bet sizes (39 BB) at 100 BB vs spec's smaller sizes
• A3 finding: 44 has unexpectedly high equity (~55%) on dry T72r vs BB's wide call range — BB's 136-class call range includes many K-high / Q-high hands that miss the board
• A4 finding: engine handles off-distribution play at 200 BB stack depth gracefully
• A5 finding: at 4 BB open, 88/99/ATs all 3-bet 100% (vs published-chart "flat"); this is consistent with the chart validation v2 finding that engine 3-bets KQs more often, and is a sizing artifact (4 BB open improves immediate odds of value-3-bet)
• AA puts substantial mass on 'limp' (25-35%) at deep stacks (80-200 BB) — Nash-valid slow-play; AA never folds at any depth (the hard sanity check)

Wall time

• 4 preflop solves × ~45s each = ~180s
• 5 analyses (equity MC) = ~40s
• Total: 217s

Test plan

• Doc generated cleanly (301 lines, within spec's 300-400)
• All five walkthroughs run with no crashes
• Engine, equity table, branch, commit SHA cited in reproduction section
• Honest divergences from spec line (A1) and from chart (A5) explicitly called out
• Companion doc cross-linked to J7o walkthrough on main