a tmp dir does not need a 2nd random var

[ReimarBauer]

Purpose of PR?:

Fixes #2536

[matrss]

This actually blocked running the test suite in a non-git checkout of the source code (e.g. extracted source archive) as it wouldn't find a commit sha then. So it is good that it gets removed.

But some changes are missing:

1. This was the only usage of the git module in tests/, so the development dependency on gitpython can be removed:

   MSS/requirements.d/development.txt

   Line 15 in a3a98d8

   gitpython

2. The docs are outdated now:

   MSS/docs/development.rst

   Lines 228 to 229 in a3a98d8

   	We have implemented demodata as data base for testing. On first call of pytest a set of demodata becomes stored
   	in a /tmp/mss* folder. If you have installed gitpython a postfix of the revision head is added.

[ReimarBauer]

1. This was the only usage of the git module in tests/, so the development dependency on gitpython can be removed:

good catch, this should have been removed there already some time longer. The move to the meta.yaml was a copy.

[matrss]

All test jobs are failing now... I suspect that to only be indirectly triggered by the changes though, because the dependency change causes a new environment resolution. I have seen the same test failures in #2298, but thought it was something specific to that PR until now. I misremembered, I think.

[ReimarBauer]

All test jobs are failing now... I suspect that to only be indirectly triggered by the changes though, because the dependency change causes a new environment resolution. I have seen the same test failures in #2298, but thought it was something specific to that PR until now. I misremembered, I think.

yeah, one timouts, 4 numpy related, #2616