* remove old deploy script

* Typo on 'M' for month in the dict

* Remove one layer of package, split in files
* Now use a unique entry point function for extraction
* Add support for multiexport (list, dict) and timedelta
* Add deploy script with a quick test

* Attribute type casting improved
* Unique entry function "extract"
* Multiexport
* Timedelta

* Now usable for GetNearest

* Simply using dtyp of asarray result in all True except for empty

* This is a quick fix, needs more work.
* Cast was always to float, which fails for boolean.
* Better handle all the type values and ask for a better cast.

* The return of the command is a tuple string, spearator ';'
* correctly convert timestamp and value

* typo naming...

Easier to debug

* Move to a single "extract" function entry point.
  It will redirect then to dedicated function for cases (Spectrum, ...)
  and for the method wanted (nearset, average, all...)

* Handle that the attribute is a list, then perform extraction for each attrivute in list

* (!) Still buggy

* For now, import pythontools only give access to core features. This will change in the future.

* Old 'self' attribute existed
* move to pandas.DataFrame return

* The previous code did not handle the retry correctly

* Fix the extractor in error message

* Multiply the number of points per the number of elements

* Add a infoattr function to retrieve the information about attribute dimension.
* When we have a 1D spectrum, switch to GetAttBetweenDates
* Serve a Pandas DataFrame

Also:
* Add Date format on helps
* Avoid running minmaxmean function

* Add __init__.py in core (empty for now)
* Add a bash source script yo load Python3 env and add the directory to pyhton path
* Add a .gitignore
* Cut out some code in ArchiveExtractor.betweenDates to outside functions