Plot bleaching estimates with correct time and/or per volume units

[talonchandler]

The current estimate-bleaching routine looks for Interval_ms metadata which is not always present.

I'll propose that we make two changes to fix this issue:

• change estimate-bleaching to read time scale with units from the .zarr metadata
• change the update-scale-metadata utility to modify the scales and units for TZYX instead of just ZYX

This will allow us enough flexibility to plot in units of "per second" or "per volume", and it does not depend on iohub correct passing to time metadata which is planned for after the universal API.

[ieivanov]

Makes sense to me

[edyoshikun]

adding to this, I just realized that a failure mode fo this is when dt gets parsed as 0.0 from the plate metadata. The default behavior should be 1.0 so that T*dt can still generate the array. Otherwise we get the following:

try:
    dt = np.float32(plate_zattrs['Summary']['Interval_ms'] / MSECS_PER_MINUTE)
except Exception as e:
    print(e)
    dt = 1
    warnings.warn(f"WARNING: missing time metadata for p={well_name}")
times = np.arange(0, T * dt, step=dt)
output_file = os.path.join(output_dirpath, well_name)
os.makedirs(output_file, exist_ok=True)

ValueError: arange: cannot compute length