fix(client): 🐛 write RMW results back into the register cache

[svasek]

🐛 What

Three client setters perform a read-modify-write against self._cached_result but never store the new value back into it. The cache is refreshed only by async_read_all, so two writes into the same shared register within one poll window make the second write start from the stale word and drop the first change.

Concrete failure: turn the hydrolysis cover-reduction flag on, then the high-temperature shutdown flag on before the next poll. Both live in the same register. The second write reads the pre-first-write value from the cache, ORs in only its own bit, and writes a word that has the first bit cleared.

🔧 Changes

• 🩹 async_set_bitmask_flag, async_set_masked_register, and async_set_filtration_speed now write the new value back into _cached_result after a successful write, so back-to-back writes stay consistent until the next full poll refreshes the cache with device truth.
• ✅ Add a regression test per affected setter asserting the cache is updated and a second consecutive write into the same register preserves the first change.

✅ Verification

• pytest 659 passed, 100% coverage
• ruff check / ruff format --check clean
• basedpyright 0 errors

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (5df0c56) to head (e97e9a9).
⚠️ Report is 1 commits behind head on main.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff            @@
##              main       #62   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            8         8           
  Lines         1222      1226    +4     
=========================================
+ Hits          1222      1226    +4     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.