à la carte standard GST reports, and exception handling for computing metrics

[rileyjmurray]

Generating a standard report can take an excessively long time. It's also possible for uncaught exceptions to be raised in the report generation process. This PR addresses both of those issues, by

1. Adding a skip_sections argument to construct_standard_report (see docs for explanation of its behavior), and
2. Adding exception handling to many (most?) modelmember metrics that might raise linear algebra errors or dimensional errors.

This PR also replaces instances of numpy.NaN with numpy.nan, for compatibility with numpy 2.0.

This splits out work from PR #410. Once this is merged into develop, I'll rebase PR #410 from that point of develop.

[coreyostrove]

All of these changes look good to me, thank you for your work, @rileyjmurray! We're asymptotically approaching the more resilient report generation we highlighted as a goal in #299. Please feel free to merge this in upon the final set of unit tests completing (assuming they have passed).