Remove `t` suffix from `Model` methods

[ricardoV94]

model.logpt, model.dlogpt and model.d2logpt would be a bit more intuitive without the t suffix. Historically that was done to distinguish the tensor graphs from the compiled functions, but those are now obtained via model.compile_logp, model.compile_dlogp and model.compile_d2logp.

[twiecki]

I never knew what the t stood for, so yes, best to remove.

[ricardoV94]

If anyone wants to take a go, this should be a simple deprecation PR:

1. Rename the new methods to pymc.model.Model without the t suffix,
2. Keep the old ones as a simple wrapper that issues a FutureWarning in between.
3. Change all the tests to make use of the new names,
4. Add some tiny tests to make sure the deprecation is being issued,
5. Create a new Github Issue to remove the deprecated names in v4.2

[cuchoi]

I can take this one if that's ok!

[twiecki]

@cuchoi Great, just open a draft PR so that people see you're working on it.

[cuchoi]

Should I update datalogpt, varlogpt, observedlogpt, and potentiallogpt, distributions.joint_logpt as well?

[ricardoV94]

Should I update datalogpt, varlogpt, observedlogpt, and potentiallogpt, distributions.joint_logpt as well?

Yes! I forgot about those

[cuchoi]

Cool. Should I update documentation such as docs/source/contributing/developer_guide.rst and docs/source/learn/core_notebooks/pymc_aesara.ipynb?

[ricardoV94]

Cool. Should I update documentation such as docs/source/contributing/developer_guide.rst and docs/source/learn/core_notebooks/pymc_aesara.ipynb?

Yes

[ricardoV94]

Opening this as a reminder to deprecate in a future major release