Remove ESPResSo system seed mechanism

[jngrad]

All thermostats are now Philox-based. The Mersenne Twister system seed infrastructure (global variables, MPI callbacks, Python bindings) is no longer needed.

[review-notebook-app]

Check out this pull request on 

You'll be able to see Jupyter notebook diff and discuss changes. Powered by ReviewNB.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (python@915df65). Click here to learn what that means.
The diff coverage is 92%.

@@           Coverage Diff            @@
##             python   #3482   +/-   ##
========================================
  Coverage          ?     87%           
========================================
  Files             ?     536           
  Lines             ?   24400           
  Branches          ?       0           
========================================
  Hits              ?   21247           
  Misses            ?    3153           
  Partials          ?       0

Impacted Files	Coverage Δ
src/core/polymer.cpp	95% <92%> (ø)

---

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 915df65...29266d1. Read the comment docs.

[jngrad]

espressomd.polymer.linear_polymer_positions has become flaky. I can reproduce the error with ctest --repeat-until-fail 1000 --output-on-failure -R polymer_linear (edit: with seed = random.randint(0,1000) to change the seed, otherwise the test always passes). If I print the Mersenne Twister seed from a failed test and re-use it in the test script, the test fails every time. The only change to polymer.cpp is that its Mersenne Twister now discards the first 1 million values to remove correlation.

[fweik]

Is there a reason why you widened the scope of mt?

[jngrad]

Nice catch, I originally moved mt out to discard the first 1 million values, but then I made a wrapper in random.hpp to avoid the situation where MT would used without warmup. I'll put it back in the capturing group.

[fweik]

The polymer generation can fail, so it should not have a random seed. Why would you expect this to work?

[jngrad]

Why would the fixed seed in the test script generate correct positions in all machines except osx? If the MT rng produces random values that are architecture-dependent, fixing the seed is useless.

[fweik]

If the MT rng produces random values that are architecture-dependent

is that so?

[fweik]

The 10000th consecutive invocation of a default-constructed std::mt19937 is required to produce the value 4123659995.

The 10000th consecutive invocation of a default-constructed std::mt19937_64 is required to produce the value 9981545732273789042

[jngrad]

I wasn't sure, that's why I used "if". I'm looking through polymer.cpp to see if something is not deterministic. The OpenGL visualizer is also currently broken on my machine, so I can't visually check the algorithm output.