fix pytorch tiledbsoma dependency

[atolopko-czi]

Fixes #955

Remove dependency upon a private method that is removed as of tiledbsoma 1.7.0

[atolopko-czi]

This fix causes performance regressions:

• 20% slower run times (10 min vs 8min)
• 3x memory usage (6GB max vs 2GB) This was tested on a filesystem-local census, querying homo_sapiens "tissue_general=='eye'", batch size 256, auto soma chunk size (7374)

The performance loss may be regained once SOMA's blockwise implementation uses its faster internal indexer (over the current Pandas indexer).

Not clear on the cause of the memory usage increase, but as users can tune memory usage with soma_chunk_size, we'll have to live with this until further effort can be put into re-tuning memory usage.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Comparison is base (23a8727) 86.17% compared to head (c49c2c5) 86.17%. Report is 1 commits behind head on main.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #957 +/- ## ======================================= Coverage 86.17% 86.17% ======================================= Files 74 74 Lines 5720 5720 ======================================= Hits 4929 4929 Misses 791 791 ``` | [Flag](https://app.codecov.io/gh/chanzuckerberg/cellxgene-census/pull/957/flags?src=pr&el=flags&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=chanzuckerberg) | Coverage Δ | | |---|---|---| | [unittests](https://app.codecov.io/gh/chanzuckerberg/cellxgene-census/pull/957/flags?src=pr&el=flag&utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=chanzuckerberg) | `86.17% <100.00%> (ø)` | | Flags with carried forward coverage won't be shown. [Click here](https://docs.codecov.io/docs/carryforward-flags?utm_medium=referral&utm_source=github&utm_content=comment&utm_campaign=pr+comments&utm_term=chanzuckerberg#carryforward-flags-in-the-pull-request-comment) to find out more.

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.