Add a ert storage endpoint for CSV_EXPORT

[frode-aarstad]

Add a ert storage endpoint for extracting the data needed by CSV_EXPORT.

Parameters:

• time_index
• column_keys

From https://github.com/equinor/semeio/blob/master/semeio/workflows/csv_export2/csv_export2.py

CSV_EXPORT2 will export selected Eclipse summary vectors to a CSV file.
The vector selection is independent of the ``SUMMARY`` keywords in the
ert config file.

The time frequency must be chosen. If ``raw``, the original timesteps from
Eclipse is chosen, and it will be individual pr. realization. If ``daily``,
``weekly``, ``monthly``  or  ``yearly`` is chosen, only data at those dates are
given for all realization. Rate data (e.g.  FOPR) is valid for the given dates,
but can not be summed up to cumulative data when time interpolation. Cumulative
columns (f.ex. FOPT) are time-interpolated linearly.

Columns are selected by a list of strings, where wildcards characters ``?``
(matches exactly one character) and ``*`` (matches zero or more characters) can
be used to select multiple columns.
Column count more than 1000 gives increased probability for problems downstream,

[TerryHannant]

Outstanding issues

• Lack of definition for what should appear in ensemble column, appears to be iteration name?
• When should time sampling be implemented? As part of this API or as separate module using this api
• Responses may not contain 1D data, how should this be mapped in output
• How to check if index column of each response is actual date type? In some example data have found other index types
• Its possible that all responses will have same index so how to combine in single flat output