Add support for optimizer checkpointing

[bwohlberg]

Add support for optimizer checkpointing. Also bump supported jaxlib/jax versions and address issues in docs.

[codecov]

Codecov Report

Attention: Patch coverage is 92.30769% with 6 lines in your changes missing coverage. Please review.

Project coverage is 93.55%. Comparing base (d4ef866) to head (2ab4a3b).
Report is 6 commits behind head on main.

Files with missing lines	Patch %	Lines
scico/optimize/_common.py	83.33%	5 Missing ⚠️
scico/numpy/util.py	95.45%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #579      +/-   ##
==========================================
- Coverage   93.58%   93.55%   -0.04%     
==========================================
  Files          92       92              
  Lines        6123     6182      +59     
==========================================
+ Hits         5730     5783      +53     
- Misses        393      399       +6     

Flag	Coverage Δ
unittests	93.55% <92.31%> (-0.04%)	⬇️

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.

[Michael-T-McCann]

Why is admm0.solve() called twice?

[bwohlberg]

The sequence is as follows:

1. Solve 5 iterations of admm0.
2. Save admm0 state.
3. Solve another 5 iterations of admm0.
4. Initialize admm1 with state of admm0 after 5 iterations.
5. Solve another 5 iterations of admm1.
6. Compare results for admm0 and admm.

The idea is that the additional 5 iterations after saving the state allow the solutions to diverge if any component of the state is not properly saved, so that the final test can just be on the solution rather than on all state components.