Update conda install scripts

[bwohlberg]

Update conda install scripts:

• Update jaxlib version
• Update astra-toolbox install method
• Default Python version now 3.9
• Handle additional dependencies for svmbir and bm3d

[codecov]

Codecov Report

Merging #178 (0ab6c9c) into main (06e4bbf) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #178   +/-   ##
=======================================
  Coverage   92.24%   92.24%           
=======================================
  Files          48       48           
  Lines        3353     3353           
=======================================
  Hits         3093     3093           
  Misses        260      260           

Flag	Coverage Δ
unittests	92.24% <ø> (ø)

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

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 06e4bbf...0ab6c9c. Read the comment docs.

[Michael-T-McCann]

In my shell (zsh on mac), ./make_conda_env.sh gives the error

./make_conda_env.sh: line 63: syntax error near unexpected token `&'

[bwohlberg]

Almost certainly due to incompatibilities between zsh and bash. How did you run it? It's intended as a bash script, and indicated as such by the initial #! /bin/bash.

[Michael-T-McCann]

How did you run it?

The same error occurs when I run it with bash make_conda_env.sh or ./make_conda_env.sh. I believe my bash version is 3.2.

[bwohlberg]

The same error occurs when I run it with bash make_conda_env.sh or ./make_conda_env.sh. I believe my bash version is 3.2.

Indeed, not incompatibility between zsh and bash, but use of an incredibly old version of bash (4.0 was released in 2009) that does not support the case fall through syntax using ;;&. I added a test for bash 4 to the script.

[Michael-T-McCann]

I can confirm the script now leads me to update bash. After doing so:

The environment I get fails a bunch of pytests due to scico.random, which in turn (I think) is due to the environment giving me jax version 0.2.27 instead of the pinned version 0.2.26.

Why doesn't the workflow catch these pytest failures on 0.2.27? Because of a bug in the workflow, see #189.

[bwohlberg]

The environment I get fails a bunch of pytests due to scico.random, which in turn (I think) is due to the environment giving me jax version 0.2.27 instead of the pinned version 0.2.26.

Fixed and tested on OSX with Python 3.9.