perf: optimize multibasis for sparse differentiation

[gdalle]

When building basis arrays:

• in the general case, assume that scalar indexing is always allowed
• add a GPUArraysCore.jl extension to handle the other case
• move OneElement code to the extension

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.93%. Comparing base (f442ed4) to head (7c86f57).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #777      +/-   ##
==========================================
+ Coverage   97.88%   97.93%   +0.05%     
==========================================
  Files         128      129       +1     
  Lines        7460     7458       -2     
==========================================
+ Hits         7302     7304       +2     
+ Misses        158      154       -4     

Flag	Coverage Δ
DI	99.02% <100.00%> (+0.07%)	⬆️
DIT	95.20% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

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