Support determistic randomness.

[jesnie]

Fixes: #1956

So, we add a seed = None parameter. We use tfp.random.sanitize_seed to transform this into a random state, in a backwards-compatible manner, and then we use tensorflow.random.stateless_* function to get randomness that only depends on our random state.

This affects our likelihoods and models. This will be a breaking change for anybody who implements custom models or likelihoods, but not for people who only call existing code.

I've also added a notebook explaining how this works.

[codecov]

Codecov Report

Base: 98.09% // Head: 98.09% // Decreases project coverage by -0.00% ⚠️

Coverage data is based on head (6206625) compared to base (42bd312).
Patch coverage: 92.53% of modified lines in pull request are covered.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #2001      +/-   ##
===========================================
- Coverage    98.09%   98.09%   -0.01%     
===========================================
  Files           93       93              
  Lines         5261     5294      +33     
===========================================
+ Hits          5161     5193      +32     
- Misses         100      101       +1     

Impacted Files	Coverage Δ
gpflow/models/cglb.py	96.10% <66.66%> (ø)
gpflow/models/sgpmc.py	95.23% <77.77%> (ø)
gpflow/models/util.py	89.13% <77.77%> (ø)
gpflow/models/gpmc.py	95.74% <80.00%> (ø)
gpflow/models/training_mixins.py	95.00% <86.66%> (-2.23%)	⬇️
gpflow/models/vgp.py	97.72% <88.88%> (ø)
gpflow/models/gplvm.py	99.24% <92.30%> (+0.04%)	⬆️
...ow/conditionals/multioutput/sample_conditionals.py	100.00% <100.00%> (ø)
gpflow/conditionals/sample_conditionals.py	95.23% <100.00%> (ø)
gpflow/conditionals/util.py	100.00% <100.00%> (ø)
... and 10 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.