bug loading `dataframe` containing a `segarray` with an `_` in column name

[21771]

Feature Description
When saving dataframes containing segarrays I have not found a straight forward way to read the data back into a dataframe. Is there a command or flags to enable automatic recreation of the segarrays from the _segments and _values arrays? This may exist but I have not found it.

**Example Behavior**
example:    df = ak.DataFrame(ak.recreate_segarrays(ak.read('foo.hdf')))
or               df = ak.DataFrame(ak.read('foo.hdf',restore_segarrays=True))
or               df = ak.DataFrame(ak.read('foo.hdf'),restore_segarrays=True)

Describe Workarounds
currently have a function to do this based on dataset names and array lengths, but am looking for a more robust capability.

[stress-tess]

@21771 i think this is a bug and segarray should be read back in by default but @Ethan-DeBandi99 would know better. I've only looked into this a bit, but it definitely seems like the code to try and rebuild the segarray is there. It looks like it's trying to read c as the column name instead of c_3 here

>>> df_dict = {"c_1": ak.arange(3, 6),"c_2": ak.arange(6, 9),"c_3": ak.segarray(ak.array([0, 9, 14]), ak.arange(20))}
>>> akdf = ak.DataFrame(df_dict)
>>> akdf.to_hdf('seg_test.h5')
>>> ak.DataFrame.load("seg_test.h5")
---------------------------------------------------------------------------
KeyError 
File ~/proj/arkouda/arkouda/dataframe.py:1612, in <dictcomp>(.0)
   1609 # update dict to contain segarrays where applicable if any exist
   1610 if len(seg_cols) > 0:
   1611     df_dict = {
-> 1612         col: SegArray.from_parts(df_dict[col + "_segments"], df_dict[col + "_values"])
   1613         if col in seg_cols
   1614         else df_dict[col]
   1615         for col in df_dict_keys
   1616     }
   1618 df = cls(df_dict)
   1619 if filetype == "HDF5":

KeyError: 'c_segments'

[stress-tess]

@21771 I think I found the issue, can you verify something for me real quick? Does the column name for the segarray in your dataframe contain any _s?

[stress-tess]

with the PR i just posted, the example I provided works

>>> df_dict = {"c_1": ak.arange(3, 6),"c_2": ak.arange(6, 9),"c_3": ak.segarray(ak.array([0, 9, 14]), ak.arange(20))}
>>> akdf = ak.DataFrame(df_dict)
>>> akdf
   c_1  c_2                          c_3
0    3    6  [0, 1, 2, 3, 4, 5, 6, 7, 8]
1    4    7          [9, 10, 11, 12, 13]
2    5    8     [14, 15, 16, 17, 18, 19] (3 rows x 3 columns)

>>> akdf.to_hdf('seg_test.h5')
>>> ak.DataFrame.load("seg_test.h5")
   c_1  c_2                          c_3
0    3    6  [0, 1, 2, 3, 4, 5, 6, 7, 8]
1    4    7          [9, 10, 11, 12, 13]
2    5    8     [14, 15, 16, 17, 18, 19] (3 rows x 3 columns)

As long as the column name of the segarray within your dataframe contains an underscore, I believe it will solve your issue