Isitfit

- enh: instead of 4 progress bars per cloudtrail region, just one for all regions

- enh: simplify cli.cost code by gathering all the mainManager setup code into `pipeline_factory` files
- enh: refactor to rename `*Listener` to `Calculator*Ec2` and `Analyzer*` to `Calculator*Redshift`
- bugfix: when redshift total > 0 but analysed = 0, do not throw exception
- enh: when iterator total=0, do not show message saying 0 resources in 0 regions
- feat: cloudtrail extended from ec2 to redshift also. Redshift cluster history now integrated into calculation of CWAU (similar to ec2, to account for changes in type/number of nodes)

- bugfix: redshift cost optimize total number analyzed was always 0. Fixed to use context variable
- bugfix: `isitfit cost optimize --n=1` was not breaking early properly. fixed
- bugfix: ec2 reporters: cleaned up usage of `n_{ec2,rc}_{total,analysed}` and `region_include`; calculating `n_*_analysed` in the analyzer; dropping `n_ec2_analysed` from mainManager
- bugfix: when datadog is not configured, the `per_ec2` listener should return the context as is, not None, thus breaking the chain for the `optimizerListener`
- enh: reporters: clean up usage of `analyzer` and `mainManager` in reporters
- enh: small test improvements

- enh: split out iterator from fetching cloudwatch data (in redshift.iterator)
- enh: use exception instead of propagating missing cloudwatch data
- enh: move cost.mainManager parts about cloudwatch to use cost.redshift.cloudwatchman. Now redshift metrics are also cached to redis
- enh: factor out datadog cached manager to a class, planning to inherit from the same base class as CloudwatchCached
- enh: move datadog caching into class that inherits from non-cached class
- enh: split out report from ec2 analyzer
- enh: split out report from ec2 optimizer
- enh: share a context dict between `per_ec2` listeners instead of passing a list of objects
- enh: move datadog out of mainManager and into listeners
- enh: move `filter_tags` and `handle_ec2obj` out of mainManager ec2 loop and into listeners
- enh: move `cloudwatchman` out of `mainManager.handle_ec2obj` into listener
- enh: move `cloudtrail` manager out of `mainManger` to listeners
- enh: move `cache_man` to listener
- enh: move `ec2_catalog` and create `ec2_common` out of the code in `mainManager` which isn't really part of the workflow
- enh: move iterator out of mainManager to cli.cost
- bugfix: ec2.reporter objects were not compliant with the new `context_all` argument format. Fixed.
- enh: redshift reports now compatible with the `context_all` argument format
- enh: (major) redshift pipelines now use the mainmanager runner, similar to the ec2 pipeline
- feat: add `--filter-region` option to `isitfit cost`

- feat: ec2 cost analyze/optimize also covers all regions
- bugfix: redis caching of missing data was checking for None, but should have checked for shape==0
- enh: add dependency on `simple_cache` for caching to local file of which regions are non-empty
- bugfix: found yet another place in the code where `tags` is None hence not iterable

- enh: split the cost.redshift.analyzer into base and derived classes for cleaner code
- enh: calculation of CWAU revised to use the average CPU, wherease the max CPU is used for the recommendations
- enh: bring back billed/used cost entries in CWAU
- enh: add integration tests to redshift cost analyze/optimize
- feat: redshift cost: cover all regions
- bugfix: redshift cost: do not multiply by 90 in analyzer since already length of dataframes = number of days
- bugfix: redshift cost: account for fractional days on first/last day