Computation of `unit_locations` doesn't respect `unit_id` order

[chrishalcrow]

The computation of unit locations (actually the bug in is compute_monopolar_triangulation) assumes that when you give unit ids, they are in the same order as the analyzer's ids.

One way to see something is wrong is that changing the order of unit_ids gives the same result:

import spikeinterface.full as si
rec, sort = si.generate_ground_truth_recording(seed=1205)
sa = si.create_sorting_analyzer(sort, rec)
sa.compute(['random_spikes', 'templates'])
unit_locs_3_6 = sa.compute('unit_locations', unit_ids = ['3','6'])
unit_locs_6_3 = sa.compute('unit_locations', unit_ids = ['6','3'])

np.all(unit_locs_3_6.get_data() == unit_locs_6_3.get_data())

>>> True

Not a problem for most workflows: but is a problem when doing merges.

Fix is pretty easy, but want to extend the AnalyzerExtensionCommonTestSuite to check for this type of bug across the repo.