[SYSTEMDS-3540] Check columns for One-Hot-Encoding before compression

[smyomous]

[SYSTEMDS-3540] Checking if column groups are One-Hot-Encoded before compression

This patch adds checks on column groups to verify if multiple columns, together, are One Hot Encoded. In that case, these column groups are compressed accordingly into an IdentityDictionary to exploit the structure of One Hot Encoded columns and optimize the subsequent operations. We observe a reduction in the number of column groups of up to 98%, with most experiments seeing a reduction of over 80%. However, the overhead on execution time is large for larger datasets due to the current implementation of the compression of One-Hot-Encoded columns.
Detailed documentation can be found in scripts/perftest/ohe_checks/README.md

[Baunsgaard]

overall good, however, some cleanup tasks left and we need to fix the tests.

[Baunsgaard]

Simply, mark the comments as resolved if you have addressed them.

[Baunsgaard]

try to look into the errors from the tests. To do this click on the failing tests

and decipher where the bug is.

One for instance is: