refactor(api): Delete dead PipetteStore code and type nozzle maps as non-Optional

[SyntaxColoring]

Overview

Minor refactors to PipetteStore to follow up on #16469 (comment) and close EXEC-768.

Test Plan and Hands on Testing

• Run something with and without partial tip configuration and make sure it doesn't raise a KeyError or anything.

Changelog

• PipetteState was typed to allow a nozzle map of None. Since #14529, this never occurs in practice—a properly-loaded pipette always has a nozzle map. So, remove the None possibility and simplify the consumers accordingly.
• As described in EXEC-768, there was a chunk of code that looked like we recently broke it. It turns out we did not break it, we made it redundant. So just delete it. See my comment below for details.
• Update PipetteStore tests to more closely reflect how LoadPipetteImplementation uses it.

Review requests

Does my comment below seem right?

Risk assessment

Low if I do the manual testing described above.

[SyntaxColoring]

Because of the if static_config: line, this chunk of code in _set_load_pipette() depended on running after _update_pipette_config(). We do not obey that order, so it looked at first like this was broken.

However, it turns out _update_pipette_config() also sets this attribute, and _update_pipette_config() is always called in practice whenever _set_load_pipette() is called. So it's moot.

For clarity, therefore, delete this code. Each attribute of self._state is now affected either by _set_load_pipette() or _update_pipette_config(), never both.

Relying on _update_pipette_config() always being called in practice whenever _set_load_pipette() is called feels a little icky to me. I think we can upgrade that from being "in practice" to being "as guaranteed by the types" by rearranging StateUpdate a little bit, but I'm not doing that here.

[sfoster1]

Definitely the way to go, thank you!