Make new Python wrapper for LOST CLI

[CullenSharp]

The Python package that provides bindings to the Lost suite is no longer maintained. Since our usage is limited to a small portion of the overall functionality, a minimal set of bindings should be able to cover this dependency.

[CullenSharp]

There were never any bindings or extensions in the traditional sense, instead the students working at UW created a Python package that wraps the LOST CLI. Its history is tracked here. I have taken most of what they wrote but changed it in some small ways: naming, path resolution, typing, exception handling.

[CullenSharp]

So far I have

• compiled LOST in a chroot'd fs
• generated a database, py-database.dat

Kind of unsure if the database file should be checked-in. It's encouraged that we generate it on the ground. My plan is to test it on the star tracker card tomorrow.

[CullenSharp]

The other thing I'm unsure about is how to bundle all the resources. I pushed a change to lost.py that utilizes importlib.resources. This is the right tool to use, but it doesn't cleanly handle the temporary directory (lost_py_wrapper/tmp) or the data directory (lost_py_wrapper/data).

[CullenSharp]

If we had proper bindings (i.e. something using pybind11, boost, python.h or similar) all of this wouldn't be an issue. I'm probably overthinking it.

Now, I'm on the fence about it, but I think we should just fork LOST and ditch UW as a predecessor completely. The repo has not seen meaningful changes in about 3 years, and it will enable use to better package it as a Python extension.

[CullenSharp]

Okay, update, I successfully ran LOST on the star tracker card and got an estimate for the dummy picture! Now it is really, really, slow. Maybe that's just the database part. Lost consumes like all the cpu for a solid few seconds.

(.venv) debian@oresat-dev:~/oresat-star-tracker-software$ python -m unittest
Narrowed catalog has 5000 stars.
Generated database with 1152972 bytes
Database flagged with 00000000000000000000000000000000
..PNG Read status: no error has occurred
Reading 1152972 bytes of database
Done
skip: mismatch prob.
Matched unique pyramid!
Expected mismatches: 6.520356e-06
Identified an additional 19 stars.
.
----------------------------------------------------------------------
Ran 3 tests in 140.036s

OK

[CullenSharp]

This is totally good to go! It's really, really, slow but it works.

I did get it to estimate, but the star-tracker-service crashes without changes to Oresat configs.