Enable use of data from EchoData object in consolidate.add_depth

[emiliom]

Overhaul consolidate.add_depth to enable using data from EchoData object. Addresses #790. See the echopype documentation for reference on several of the variables.

Rotations are performed using scipy.spatial.transform.Rotation.

Notes and caveats:

• See the TODOs in the code.
• If both an EchoData object and external data (as scalar arguments) are passed, only data from the EchoData object is used.
• I'm seeing too many nans in the resulting depth variable for the sample EK80 file, so I suspect something isn't working properly.
• AZFP tilt_x/y variables are not handled. Only standard SONAR-netCDF4 variables are used. It assumes that tilt_x/y variables have already been transformed into pitch & roll. I'll create a new issue for this.
• I learned recently (from the EK80 reference manual) that EK80 TransducerAlphaX/Y/Z variables are in degrees. We're currently interpreting them as a unit vector when mapping them to beam_direction_x/y/z. A conversion will need to be implemented in set_groups_ek80 (and ek60?). I'll create a new issue for this. For the sample file (below), it's not an issue b/c all TransducerAlphaX/Y/Z values are zero. I couldn't find an EK80 file with TransducerAlphaX/Y/Z values that are not all zero.
• We don't have a sample EK60/80 raw file that has all relevant variables populated with what appear to be valid values. The one file I'm using, from 2021 Hake survey ("ek80/ncei-wcsd/SH2106/EK80/Reduced_Hake-D20210701-T131621.raw"), only has some of the variables with apparently valid, non zero data: pitch, roll, and vertical_offset. vertical_offset (based on heave) is also suspicious b/c values are close to zero. water_level (0) + vertical_offset is supposed to be the positive distance from the ship origin.

[codecov-commenter]

Codecov Report

Merging #1205 (43e44cd) into dev (9dfe5ba) will increase coverage by 15.77%.
Report is 3 commits behind head on dev.
The diff coverage is 97.61%.

@@             Coverage Diff             @@
##              dev    #1205       +/-   ##
===========================================
+ Coverage   77.08%   92.85%   +15.77%     
===========================================
  Files          67        3       -64     
  Lines        5921      196     -5725     
===========================================
- Hits         4564      182     -4382     
+ Misses       1357       14     -1343     

Flag	Coverage Δ
unittests	92.85% <97.61%> (+15.77%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files	Coverage Δ
echopype/consolidate/api.py	94.95% <97.61%> (+1.13%)	⬆️

... and 64 files with indirect coverage changes

📣 Codecov offers a browser extension for seamless coverage viewing on GitHub. Try it in Chrome or Firefox today!

[emiliom]

Notes to self for creating mock EchoData data for testing the new functionality. Here's the minimal code needed to create a bare EchoData object that has just the elements needed by consolidate.add_depth: a sonar_model attribute and the "Platform" and "Sonar/Beam_group1" groups.

from datatree import DataTree
import xarray as xr
from echopype.echodata.echodata import EchoData

ed = EchoData(sonar_model="EK80")

ds1 = xr.Dataset(dict(bar=("x", [1, 2])))
ds2 = xr.Dataset(dict(bar=("x", [10, 20])))
tree = DataTree.from_dict({'Platform': ds1, 'Sonar/Beam_group1': ds2}, name='root')
ed._set_tree(tree)

[emiliom]

Added a test function to generate mock EchoData object with only the variables (and one attribute) required by consolidate.add_depth. This function partially populates only the Platform, Sonar and Sonar/Beam_group1 groups, plus root by default. Also added a test (test_add_depth_from_echodata_mock) that uses this mock data, together with a mock Sv Dataset.

[leewujung]

@ctuguinay : This is also related to #1152. Let's talk through this.

[leewujung]

We'll close this now because it'll be superseded by #1319 .