Prevent Joy-Con pointer jumps when surface tracking resumes

[jmonster]

Joy-Con optical reacquisition fix

Head: 237f7b5fc9042c0552d7b873bef944251bf4b2ec. Ready for review; all seven workflows for this unchanged head passed. Draft status has been removed. No PR merge or force-push was performed.

The opt-in dashboard mouse adapter retained a primed optical counter baseline through lift-off, out-of-range lift distance and unusable surface quality. It also primed on an initial no-surface report. A following valid report could turn an untracked counter jump into cursor movement.

Invalidate the baseline and fractional residue on rejected surface reports. The first usable report only rearms; subsequent reports retain modulo-65536 delta conversion. Both Joy-Con halves are covered. Button ownership, valid tracked movement, permission behavior and existing test assertions are retained.

Only Sources/Switch2KitApp/Output/MouseController.swift and its boundary tests change. No controller protocol, Swift/C ABI, native SDL/emulator, signing, packaging or permission change.

Exact-head CI

Workflow	Result
macOS package/regressions/app/demo	PASS
Swift source/native C distribution	PASS
Real SDL3 native consumers	PASS
Swift concurrency	PASS
SDL regressions	PASS
Packaged runtime	PASS
Emulator integration	PASS

These are the workflows that ran for the unchanged head, not a claim that the newer Linux workflow ran on this older branch. The completion audit found no review comments, submitted change-request reviews or unresolved review threads.

Validation

The original regression failed against the preceding production source and passed with the fix. Coverage includes both Joy-Con halves, all three rejection conditions, initial contact, counter jumps/wrap and clearing fractional residue. Complete retained regressions, package build, 82 debug and 82 release package tests with warnings as errors, and whitespace checks passed in the original local run. The downloaded CI source archive matches the published head's complete tree and executable modes.

Additional completion validation combined current main a187693b8ced4328758c4e242a4eb203d26608a2 with #63, #64, corrected #65 (4fef3dc) and corrected #68 (9757016). The local combined tree is fa80b2e7eab3dfadc3196e73c01c9d59079b1422; it was not published or merged.

• Complete retained regression suite: PASS, including the pointer-boundary tests and 20 isolated BlueZ scenarios.
• Package tests with warnings as errors: 92 debug and 92 release PASS.
• Real SDL 3.4.16/native host suite, including pinned Cemu and Dolphin motion processors: 8/8 PASS.
• Native facade relocation with lib and lib64, two real C++ consumers and missing-library negative controls: PASS.
• Forty additional release BlueZ transition cases and a separate all-four-model BlueZ → actual C ABI/SDL → pinned Cemu motion harness: PASS. These are additional local checks, not new committed CI tests or physical measurements.

Pointer tests compile production logic with fake event creation/posting. They do not establish actual desktop pointer behavior or physical optical quality. The cross-PR tests also use synthetic controller data/profiles; physical Bluetooth/gameplay acceptance remains separate.