Automatic slurm allocations

[MilesCranmer]

This adds the cluster_manager="slurm_native" option which uses https://github.com/kleinhenz/SlurmClusterManager.jl. The idea with this is that you could potentially just have

sbatch -N 2 --ntasks-per-node=64 search.py

and it will "just work" and set up the search over the cluster. This is the much more standard way to use slurm. This will get the right slurm configurations from the environment variables.

TODO:

• Do a multi-node test. Maybe @wkharold would be interested in testing too?
• Set up heap size hint We don't actually need to do this! Since slurm will partition the memory for us and each Julia process will be aware.
• Add a detailed example to docs for how to set this up.

[coveralls]

Pull Request Test Coverage Report for Build 12581536478

Details

• 11 of 25 (44.0%) changed or added relevant lines in 3 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.7%) to 92.706%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
pysr/sr.py	0	1	0.0%
pysr/julia_extensions.py	9	14	64.29%
pysr/julia_helpers.py	2	10	20.0%

Totals
Change from base Build 12512611011:	-0.7%
Covered Lines:	1398
Relevant Lines:	1508

---

💛 - Coveralls