Refactor tests with fixtures

[jnsbck]

This PR adds test fixtures for the modules that have to be re initialized again and again during testing. This should reduce overhead and make tests more readable. The fixtures are such that if i.e. Branch(nseg=4) was already used by a prev test, the fixture returns a copy of it, if not it is build from scratch and cached. Higher order modules, i.e. Cell([Branch(nseg=4)], [-1,0,0]) that are build later can also make use of the pre-build submodules via the fixtures.

This compliments #479 and contributes towards #449.

[jnsbck]

PR is currently WIP. Might also try to reduce test runtime.

[jnsbck]

PR introduces simpler test primitives, adds slow markers for tests that take longer than 10s to run and also standardizes the test-cases a bit more, see conftest.py for fixtures. Also reduced the simulation times as much as possible by reducing t_max in the stimuli. However, I can see no major speed improvements. Feedback would be greatly appreciated.

The following I am unsure about:

• Should we shorten integration time of some hardcoded testcases, i.e.

  jaxley/tests/jaxley_identical/test_basic_modules.py

  Line 26 in 32068e9

  def test_compartment(voltage_solver: str):

• Should we standardize test currents even more? I.e. short_pulse, long_pulse, dummy
• I noticed while playing around with the currents that test_type_optimizer is extremely brittle w.r.t. changing t_max. Any ideas why? Could we make it more robust?

EDIT: These are the currently used stimuli.

• 24: i_delay=0.1, i_dur=1.0, i_amp=0.1, t_max=2.0
• 8: i_delay=0.5, i_dur=1.0, i_amp=0.02, t_max=5.0 # -> shorten?
• 3: i_delay=0.0, i_dur=0.0, i_amp=0.0, t_max=0.1 # dummy current
• 2: i_delay=0.5, i_dur=0.5, i_amp=0.1, t_max=10.0 # -> shorten?
• 2: i_delay=0.5, i_dur=1.0, i_amp=0.2, t_max=5.0 # -> shorten?
• 1: i_delay=0.1, i_dur=1.0, i_amp=0.1, t_max=5.0 # -> brittle test?
• 1: i_delay=0.5, i_dur=1.0, i_amp=0.1, t_max=5.0 # -> shorten?
• 1: i_delay=0.5, i_dur=1.0, i_amp=2.0, t_max=10.0 # -> shorten?

[michaeldeistler]

Amazing!