benchmarking

The original issue was reported in dotnet/performance#1701:

// Benchmark Process Environment Information:
// Runtime=.NET 6.0.0 (6.0.21.11801), X64 RyuJIT
// GC=Concurrent Workstation
// Job: Job-AQTFSE(PowerPlanMode=00000000-0000-0000-0000-000000000000, IterationTime=250.0000 ms, MaxIterationCount=20, MinIterationCount=15, WarmupCount=1)

OverheadJitting 

For different use cases, like bencheeorg/benchee_html#10 it'd be great to have statistics about statistics - what I'd call "meta statistics" - although there's probably some better real statistics name for this :)

What should be in there (that I know of so far):

• job size (how many jobs are in there)
• minimum of run times over all jobs
• maximum of run times over all jobs

This should be

By exploiting the callbacks before_forward and similar it is very easy to log samples of the training dataset, number of patterns, features statistics and other metrics.

We can discuss in here what it could be useful to add.

There seem to be some vulnerabilities in our code that might fail easily. I suggest adding more unit tests for the following:

• Custom agents (there's only VPG and PPO on CartPole-v0 as of now. We should preferably add more to cover discrete-offpolicy, continuous-offpolicy and continuous-onpolicy)
• Evaluation for the Bandits and Classical agents
• Testing of convergence of agents as proposed i