[Bugfix] Re-allow non-integral line/grid qubits

[daxfohl]

Fixes #7102. No longer coerces x/row/col to int. Wraps hash expression with hash instead.

On my laptop it's also a tiny bit faster. It creates 10,000,000 LineQubits in 2.1s; old code took 2.5s.

Also, note that I left all the type annotations int, which is what they have always been. I looked into changing these into numbers.Rational or SupportsFloat or something, but it's too invasive. Lots of code seems to assume integral x/row/col values, and mypy blows up when you try to expand it.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 98.15%. Comparing base (2ad4136) to head (25c8ab0).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #7110      +/-   ##
==========================================
- Coverage   98.15%   98.15%   -0.01%     
==========================================
  Files        1089     1089              
  Lines       95251    95255       +4     
==========================================
- Hits        93497    93495       -2     
- Misses       1754     1760       +6     

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

[dstrain115]

Thanks for fixing this!

[pavoljuhas]

LGTM with a couple of small comments. Thank you!