adding error messages and check for NaN values to compiled_dynamics

[sabinala]

Adds a descriptive error message when an underflow error is thrown by an adaptive step size solver, or infinite/NaN values are returned for states and/or observables.

See #643.

[sabinala]

The changes to compiled_dynamics have been tested (and shown to work for the simulate and calibrate interfaces, as well as the ensemble versions) in this notebook: https://github.com/ciemss/pyciemss/blob/sa-solver-error-msgs/docs/source/Solver_error_messages.ipynb.

Normally, I would write a unit test for this, but https://github.com/DARPA-ASKEM/simulation-integration has been archived, so I can no longer add models to this repo! @SamWitty let me know what you think, either we can merge the PR without a test, or I can add the test and add the offending model to another location and then to tests/fixtures.

[sabinala]

Also, I do not know why linting is failing! I ran make format before committing, and now I'm getting this error message from lint:

32s
Run ./scripts/lint.sh
+ mypy --ignore-missing-imports pyciemss/
pyciemss/compiled_dynamics.py:267: note: By default the bodies of untyped functions are not checked, consider using --check-untyped-defs  [annotation-unchecked]
pyciemss/compiled_dynamics.py:268: note: By default the bodies of untyped functions are not checked, consider using --check-untyped-defs  [annotation-unchecked]
Success: no issues found in 30 source files
+ isort --check --profile black --diff pyciemss/ tests/
+ black --check pyciemss/ tests/
would reformat /home/runner/work/pyciemss/pyciemss/pyciemss/integration_utils/intervention_builder.py

Oh no! 💥 💔 💥
1 file would be reformatted, 42 files would be left unchanged.
Error: Process completed with exit code 1.

I have not touched intervention_builder.py, the only file that is touched in this PR is compiled_dynamics.py

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[SamWitty]

lgtm!