fix: reject over-large sliding window (raise instead of silent-empty)

[breimanntools]

Resolves the deferred sliding-window guard from the July 2026 correctness audit.

check_match_slide_start_slide_stop_window_size had a typo — min_window_size = slide_stop - slide_stop (always 0) — so the guard if window_size < min_window_size could never fire. Requesting a sliding window larger than the explicit slide span therefore silently returned an empty list. The guard now raises a clear ValueError when window_size > slide_stop - slide_start + 1 (the point past which no window fits, per the backend's n_windows formula).

This is a behavior/contract change (silent-empty → raise), decided in favor of raising. The two hypothesis tests that encoded the old silent-empty behavior are updated to the new contract, and a dedicated over-large-window test is added.

Only fires when slide_stop is passed explicitly (the default slide_stop=None path is unchanged); no internal callers relied on the silent-empty behavior.

Part of #342.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.85%. Comparing base (dc4289c) to head (ae74ecd).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #350      +/-   ##
==========================================
+ Coverage   94.84%   94.85%   +0.01%     
==========================================
  Files         196      196              
  Lines       18783    18783              
  Branches     3181     3181              
==========================================
+ Hits        17815    17817       +2     
+ Misses        632      631       -1     
+ Partials      336      335       -1     

Files with missing lines	Coverage Δ
aaanalysis/data_handling/_seq_preproc.py	100.00% <100.00%> (+1.62%)	⬆️

Components	Coverage Δ
cpp_core	94.95% <ø> (ø)

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.