Pandas "shell" as alternative to `dolt sql`

[addisonklinke]

Many data scientists and machine learning engineers are more comfortable with Pandas over raw SQL. What do folks think about adding a new dolt pd command to the CLI options? In my proposal, it would work like this

$ dolt pd -h
NAME
	dolt pd - Load SQL table into an interative Pandas dataframe

SYNOPSIS
	dolt pd [-v <version>] <table> [<repo>]

OPTIONS
	<table>
	  The existing table to read from.
	<repo>
	  Path to local Dolt repository (default: current directory).
	-v <version>
	  Specific version of the table to read using Dolt's AS_OF clause (default: latest).

$ dolt pd -v branch my_table /some/dolt/clone

# This chunk is automatically executed to save users the boilerplate
# Raise an error if the required Python packages are not installed (or distribute them with the dolt binary)
>>> import doltcli
>>> from doltpy.cli.read import read_pandas
>>> db = doltcli.Dolt('/some/dolt/clone')
>>> df = read_pandas(db, 'my_table', as_of='branch')```

# Then you're free to do various adhoc EDA/ETL tasks
>>> df.head()
>>> print(df.some_column.value_counts())

It'd be nice to default to iPython if it's available under the current environment

[timsehn]

This is a really cool idea. Let us discuss.

[timsehn]

I don't think we're ever going to get to this. Closing.