feat(mlflow): autolog support for BassModel

[anevolbap]

Closes #2588

Adds Bass model support to pymc_marketing.mlflow.autolog, following the existing MMM and CLV patterns:

• New log_bass flag (default True) patches BassModel.fit to log the model configuration, stamp the run id on idata.attrs, and store the idata.nc artifact.
• New log_bass_configuration helper logs model type, version, sampler config, and the prior configuration (m, p, q, likelihood).
• Module and autolog docstrings document the new integration with an example.

While testing this, autolog on a default-prior BassModel failed about half the time with ValueError: lam < 0 or lam contains NaNs. The cause is pm.model_to_graphviz, which draws from the prior graph to evaluate plate shapes; the default m ~ Normal(0, 10) prior then produces a negative Poisson rate. The same happens with pm.sample_prior_predictive on a default BassModel, so the root cause is the prior itself: should the default prior for m be positive (HalfNormal, LogNormal, truncated Normal)? Happy to open a separate issue for that. Meanwhile, a separate commit makes log_model_graph treat graph creation as best-effort, logging a message and continuing, the same way it already handles a missing graphviz install or a render failure.

Sampler diagnostics, ArviZ summary, model graph, and metadata logging already worked through the patched pm.sample that BassModel.fit calls internally; this PR only adds the Bass-specific layer described in the issue.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.20%. Comparing base (649547e) to head (12e0851).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2624   +/-   ##
=======================================
  Coverage   94.19%   94.20%           
=======================================
  Files         100      100           
  Lines       15384    15403   +19     
=======================================
+ Hits        14491    14510   +19     
  Misses        893      893           

☔ View full report in Codecov by Harness.
📢 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.