[FIX] CSV Import: Add explicit datetime conversion

[ales-erjavec]

Issue

Fixes an error with dev version of pandas as seen on CI i.e.

FAIL: test_load_csv (Orange.widgets.data.tests.test_owcsvimport.TestUtils.test_load_csv)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/runner/work/orange3/orange3/.tox/beta/lib/python3.11/site-packages/Orange/widgets/data/tests/test_owcsvimport.py", line 449, in test_load_csv
    self.assertSequenceEqual(
AssertionError: Sequences differ: [dtype('O'), dtype('float64'), dtype('O'), Catego[100 chars]64')] != [dtype('<M8[ns]'), dtype('float64'), dtype('O'), [24 chars]64')]

First differing element 0:
dtype('O')
dtype('<M8[ns]')

+ [dtype('<M8[ns]'), dtype('float64'), dtype('O'), 'category', dtype('float64')]
- [dtype('O'),
-  dtype('float64'),
-  dtype('O'),
-  CategoricalDtype(categories=['one', 'three', 'two'], ordered=False, categories_dtype=object),
-  dtype('float64')]

Description of changes

Add explicit to_datetime conversion for requested datetime columns when read_csv failed to convert all values.
Also remove obsolete workaround float precision parsing.

Includes

• Code changes
• Tests
• Documentation

[codecov]

Codecov Report

Merging #6696 (0bed791) into master (7b68883) will decrease coverage by 0.02%.
Report is 10 commits behind head on master.
The diff coverage is 75.00%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #6696      +/-   ##
==========================================
- Coverage   88.08%   88.06%   -0.02%     
==========================================
  Files         322      322              
  Lines       70105    70257     +152     
==========================================
+ Hits        61755    61875     +120     
- Misses       8350     8382      +32