Implement three-part-make for the EphysRecording table

[ttngu207]

Refactoring and modularization of data processing workflow:

• Split the make method in EphysRecording into three methods: make_fetch, make_compute, and make_insert, each responsible for fetching data, computing results, and inserting records, respectively. This improves code clarity and separation of concerns. [1] [2]
• Updated usage of supported acquisition software and probe types to fetch them once and pass them through the workflow, reducing redundant database queries.

Dependency management:

• Changed the installation of spikeinterface in setup.py to use a direct GitHub reference, ensuring the latest version is used.

Probe model identification improvement:

• Modified the fallback assignment for unknown probe models in the SpikeGLX reader to use self.probe_PN instead of str(probe_model), providing more accurate probe identification.