Minor changes to the storage of simulation results

[gregoryross]

Details

• The recorder class now saves the log acceptance probability and number of attempts. Saltswap.Swapper has been modified to allow the former.
• The cumulative protocol work was being stored in Swapper as variables in numerous different ways. As these have now been superseded by the recording of the proposal and associated cumulative work, the superfluous variables have now been removed
• The test for the reading and writing of NetCDF storage file now removes the temporary file that gets created during for the test.

[gregoryross]

Thanks for looking through this @bas-rustenburg.

Let me know if/how your test passes though.

My tests are passing.

Before merging, I'll now simplify the dimensions of the arrays I'm storing and switch to storing data in 32 bit following suggestions by @jchoderas. My preliminary simulations are producing large NetCDF files and I don't think I need 64 bit precision with the variables I'm storing.

[jchodera]

In case you wanted more info on chunksize.

[gregoryross]

Thanks @jchodera