Add Opening (RFI) range viewer alongside push/fold

[testtest126]

Summary

Adds standard-stack (20–100bb) opening / raise-first-in ranges, as a second mode on the existing "Preflop Ranges" grid — the biggest gap identified in a read-only assessment of the app for a beginner MTT/PKO player: the only existing preflop model was short-stack push/fold (1–20bb), covering a minority of an MTT's actual preflop decisions.

⚠️ Draft — poker content needs your sanity check before this is canon. See "Correctness — please review" below before merging.

• PokerKit/Sources/PokerKit/OpeningRange.swift — new model, mirrors PushFoldRange.swift's shape exactly (same breakpoint/interpolation pattern, same ChenScore-based hand ranking, reuses PushFoldRange.scoreThreshold(forPercentage:) directly rather than a second hand-ranking system).
• PreflopGrid.swift — one new function, openingDecisions(position:effectiveStackBB:), alongside the existing decisions(...). No duplication of the 169-hand grid enumeration.
• PreflopRangeView.swift — a "Push/Fold" / "Opening" segmented control that swaps the model and the stack-slider range (1–20bb vs 20–100bb).
• StudyTool.swift + README.md — fixed a pre-existing honesty gap: the home-screen summary text promised "opening/3-bet/4-bet ranges" that didn't exist; now accurately describes what's built (opening + push/fold; still no 3-bet/4-bet).
• Tests: OpeningRangeTests.swift (9 tests) + 2 new tests in PreflopGridTests.swift — AA always opens, ranges widen monotonically by position and as the stack shortens, button > UTG, SB > BTN, the load-bearing "grid matches direct decision" invariant, interpolation/clamping. 94/94 PokerKit tests pass (83 existing + 11 new).

Correctness — please review

Per this repo's own rule (bad poker math is worse than none), the full source basis and every extrapolation is disclosed in ai-docs/RANGES.md, not just asserted. Short version:

• 100bb column is sourced: PokerCoaching.com's "Implementable GTO Charts" (6-max, mixes removed), cross-verified by fetching it twice independently — UTG 10.1%, LJ 17.6%, HJ 21.4%, CO 27.8%, BTN 43.5%, SB 62.3%.
• SB is deliberately tightened: a second source (freebetrange.com) put SB at 39–47%, meaningfully lower than the 62.3% GTO figure above. Rather than pick whichever number sounds more authoritative, the shipped table uses 45% — the conservative end — and this is called out explicitly as a disclosed judgment call, not a citation.
• 40bb column: extrapolated from a single anchor point (UTG ≈13% from a separate source, a +3pt widening vs. the 100bb UTG figure), applied uniformly. Not independently sourced per-position.
• 20bb column: hand-tuned extrapolation with no direct source (+6pt vs. 100bb), existing mainly so the slider has a smooth floor near where PushFoldRange picks up. Flagged in the doc as the least trustworthy column.

Please specifically sanity-check the SB override and the 40bb/20bb extrapolations — those are the parts I'm least confident in and where your own judgment/experience is worth more than a second web search.

Test plan

• swift test --package-path PokerKit — 94/94 passing
• xcodebuild build -scheme Poker (iPhone 16 Pro simulator) — BUILD SUCCEEDED
• Manual verification in simulator: mode toggle changes the grid/summary/legend, stack slider re-ranges and resets to a sensible default (50bb) on mode switch, BTN selection widens the range as expected (UTG 14% → BTN 50% at 50bb)
• Owner sanity-check of the range numbers themselves before merge

🤖 Generated with Claude Code