adds more stringent dry run

[fredshone]

adds a more careful init check to the csv comparison benchmark tool. More generally adds a "dry-run" build method to base factory.Tool and Workstation classes. This is then called during the (existing) dry run command, eg:

elara run example_configs/benchmarks.toml -d

Generally i am trying to give cheap and early guarantee that everything will work,

I am not including checking of input paths to these dry-runs. Because it is intended to be called in BitSim, prior to input paths being known. Behaviour is only tested very basically but also added to smoke tests. I have confirmed expected behaviour via some local experiments.

the python-levenstein warning is also now suppressed. This Closes Issue 211

[fredshone]

baaa - i could also have just put this in the init method... will change

[fredshone]

description updated