Docker image preparation and bump COMPASS version for cal/val delivery

[seongsujeong]

• SAS Docker images based on develop version of ISCE3
  • Specfile for the Docker image above
• Bump the COMPASS version from 0.1.5 to 0.2.0

NOTE:

• The Dockerfile used for the last Gamma release is still in place, in case the new version of ISCE3 becomes available
• This is a PR to prepare a CSLC-S1 SAS as a Docker image. ISCE3 is installed by building the source code from the most up-to-date commit in develop branch in internal GitHub.
• The ISCE used in this Docker image will be replaced either tagged commit (preferably mirrored to the public GitHub), or condo package of ISCE3, once either of them becomes available.

TODO:

• [x] Optimize docker image size

[seongsujeong]

@LiangJYu @vbrancat As you aware, ISCE3 0.13.0 has become available this morning. The Docker image that I've prepared is based on the conda package of the ISCE3.

Therefore, this PR will discard the Dockerfile.isce3.dev based on the on-the-fly ISCE3 build, and focus on updating the specfile for cal/val SAS delivery.

[seongsujeong]

@LiangJYu Can you take a look if the updates in the Docker image is looking good? Thanks.

[seongsujeong]

A note about CI test failure:

Did a bit of modification on test_qa.py to print out the stats written in the HDF5, and found this information:

Testing: mean of power = 2552.599609375 Testing: min of power = 2.686270818230696e-05 Testing: max of power = 56852836.0 Testing: mean of phase = -0.0004393019189592451 Testing: min of phase = -3.1415927410125732

The CI test failed when testing the min. value of phase. Apparently its absolute value is slightly larger than np.pi The value would have been calculated using np.nammin(), which looks okay to me.

Running pytest locally in aurora did not have such issue.

As a temporary fix, I will put a tolerance of 1.0e-8 when checking the stats from phase.