Rework MVBS tests

[leewujung]

Right now our tests for compute_MVBS are pretty messy and we can benefit from reworking this for robustness and readability.

A laundry list of the few things I noticed while working on #753 were:

• operations in _construct_MVBS_toy_data and _construct_MVBS_test_data either are hard to decipher or seem can be simplified. It may be the best to rewrite these.
• factor out common components between test_compute_MVBS_index_binning and test_compute_MVBS to reduce redundancy and avoid error
• the above work with constructed data
• test_preprocess_mvbs runs through the mechanism for real data, we should keep this but consider adding variations of the target echo_range and ping_time binning intervals
• add test for Sv data with different length along the range_sample dimension for different channel
• use np.equal instead of np.allclose when testing computed values
• use pytest parameterization to reduce clutter and redundancy
• separate out MVBS-related tests to a separate .py file

[lsetiawan]

This may be addressed in #1124