Improve cli options for use w/o a pipeline file

[cwensel]

What additional options/features are needed to make Tess more useful standalone on the command line:

• Write all fields to a json file, not just a single field - Needs a new Schema or embed toJson intrinsic
• --input-fields so only the required columns are read from parquet, ensure type information is honored
• --input-format so a directory of files of the same format can be read without seeking the first filename
• --transform to allow chaining transformations/filters
• --output-limit-head & --output-limit-tail option to limit the number of records emitted
• --output-embed-schema to print the header information on write