[BUG] Allow nulls in partition column

[colin-ho]

Closes #2292

Currently Daft panics if there are nulls in a partition column, the detailed error message can be found in the linked issue.

A simple reproduction:

from deltalake import write_deltalake
import pandas as pd
import daft

df = pd.DataFrame(
    {
        "group": [1, 2, 3, None],
        "num": list(range(4)),
    }
)
write_deltalake("z", df, partition_by="group", mode="overwrite")

df = daft.read_deltalake("z")
df.show()

This PR modifies the partition spec equality logic and partition pruning semantics to allow reading nulls in partition columns.

[codecov]

Codecov Report

Attention: Patch coverage is 81.81818% with 2 lines in your changes missing coverage. Please review.

Please upload report for BASE (main@408f977). Learn more about missing BASE report.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #2344   +/-   ##
=======================================
  Coverage        ?   79.00%           
=======================================
  Files           ?      475           
  Lines           ?    55264           
  Branches        ?        0           
=======================================
  Hits            ?    43661           
  Misses          ?    11603           
  Partials        ?        0           

Files	Coverage Δ
src/daft-scan/src/python.rs	67.48% <100.00%> (ø)
src/daft-stats/src/partition_spec.rs	78.94% <77.77%> (ø)

[samster25]

So for partition spec we want the behavior that null == null? if so, we should document that here

[samster25]

a simpler way to represent this could be:

else if self_is_null xor other_is_null {
  return false;
}