Shmessy

- Improve inference time by removing non-relevant date-time combinations
- Date only patterns list (Do not attach timestamp combinations to these dates)
- Do not attach timestamp combinations to static date formats
- New dynamic + static date/timestamp patterns:

"%Y-%m-%dT%H:%MZ", 2024-01-20T11:30Z
"%B %d, %Y", January 23, 2024
"%b %d, %Y", Feb 17, 1995
"%d %b %y, %a", 10 Apr 24, Wed

["%Y", "%m"], 2022-07 | 2022 07 | 2022/07 | 2022.07
["%Y", "%b"], 2022-Jul | 2022 Jul | 2022/Jul | 2022.Jul
["%m", "%y"], 07-22 | 07 22 | 07/22 | 07.22
["%b", "%y"], Jul-22 | Jul 22 | Jul/22 | Jul.22
["%b", "%Y"], Jul-2022 | Jul 2022 | Jul/2022 | Jul.2022

- Set min/max boundaries to integer to be the values of BIGINT (+-9,223,372,036,854,775,807)
- Enforce casting the float value to np.Float64 instead of any numeric value

Major release. We broke the API in order to make it much more simple.

- All of the configurations & flags pushed to the constructor with default values.
- Date / Datetime bug fix that improve the infer duration - do not try to validate all of the sample data after the first mismatch.
- Create dynamic patterns using much more combinations than before for Date / Datetime types.

- Improve datetime speed recognition
- create datetime utils for core readabillity
- Dynamic date types

New date format:

- %b-%d-%Y

New datetime format:

* %d/%m/%y %H:%M