[heli-os]

Summary

Promote compact token counts when one-decimal rounding would produce 1000K or 1000M. For example, 999,950 now displays 1M, and 999,999,999 displays 1B. Existing precision, trimmed zeros, and negative signs are preserved; cost formatting is unchanged.

The unit was previously selected using the unrounded magnitude. Existing tests covered exact unit boundaries and ordinary values, but not values that round across a boundary. The new regression test covers immediately below/at the rounding thresholds, immediately below/at exact unit boundaries, and negative counterparts. A call-site sweep confirmed menu, usage, growth, and shop displays share this formatter; there is no duplicate token abbreviation implementation to update.

Rounding thresholds

K and M use one decimal place. The rollover therefore starts when the scaled value reaches 999.95, not when the raw token count reaches the next power of 1,000:

• 999,949 / 1,000 = 999.949 → 999.9K.
• 999,950 / 1,000 = 999.95 → 1000.0K → 1M.
• 999,950,000 / 1,000,000 = 999.95 → 1000.0M → 1B.

Using 999,999 or 999,999,999 as the threshold would still leave earlier values displaying 1000K or 1000M. The fix keeps rounding; it does not truncate counts.

Comparison rendered from the actual Swift formatter outputs at the base and PR commits; this is a formatter-output table, not an app screenshot.

Type of change

• Bug fix
• New feature
• Refactor / cleanup
• Documentation
• Other:

Validation

• swift build passed locally.
• Ran the actual formatter source in a standalone Swift harness: 30 checks, 8 failures before the fix and 0 after it.
• Inspected instrumented llvm-cov show --show-regions output for the formatter with the same harness.
• swift test --filter TokenFormatterTests and ./scripts/test-gate.sh were attempted, but this Mac has only Command Line Tools and cannot import XCTest. The standalone harness is not a substitute for the full test suite; CI should run the committed XCTest regression.
• GitHub CI passed build-test and secret-scan: 1,381 tests executed, 14 skipped, 0 failures; logic-core line coverage 93.27%. The new boundary regression passed. Run: https://github.com/chattymin/PokeTokenBar/actions/runs/36311970875
• git diff --check passed.

Checklist

• swift build and swift test pass locally
• PR title and description are written in English
• No copyrighted assets, secrets, or private tooling references are committed
• Tests were added or updated for this change