MemoryError issues during deploy

[virtuald]

The working theory is that pip writes the wheels to a temporary directory, which fills RAM, which causes MemoryErrors to occur. We haven't definitively measured this, but some ideas to address this involve:

• Kill the NI webserver during deploy
• Kill any user program during deploy
• Use NFS instead of a webserver to provide pip the artifacts? https://pypi.org/project/ShenanigaNFS/
• Use a FUSE filesystem to provide pip the artifacts

[virtuald]

We kill the NI webserver and the user program now, maybe things are better now?

[virtuald]

Nobody has complained about this recently, so it's probably fine for now?