Improve output file management

[davidslater]

Currently, when output_dir or output_filename are set in the config, they are used as part of the path for the output results file. However, it is not very useful for collating files as timestamps and IDs are still appended. So, if I set output_dir="hi_there" and output_filename="hi5.json", I end up with something like this: an output directory ~/.armory/outputs/hi_there_2022-03-17T174050.556733 and output file hi5.json_1647538856.json.

For programmatic parallelization and collation of results, it would be much nicer to have them write to ~/.armory/outputs/hi_there/hi5.json. More control would also be nice, so that you can drop the output file into the cwd, not just into ~/.armory/outputs.