Use native zarr for trace worker instead of xarray for memory efficiency.

[BrianMichell]

When scaling segy ingestions we observed higher than expected memory usage. I believe this is in part due to unavoidable overhead in the Xarray API internals. For that reason I have begun to strip out xarray API calls in favor of zarr.

This appears to have helped reduce large scale memory usage, although it is still consuming more memory than expected.

[codecov]

Codecov Report

❌ Patch coverage is 14.28571% with 18 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.87%. Comparing base (c3ba558) to head (104507c).
⚠️ Report is 191 commits behind head on main.

Files with missing lines	Patch %	Lines
src/mdio/segy/_workers.py	10.00%	18 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #710      +/-   ##
==========================================
+ Coverage   85.30%   88.87%   +3.56%     
==========================================
  Files          46       84      +38     
  Lines        2219     4529    +2310     
  Branches      306      294      -12     
==========================================
+ Hits         1893     4025    +2132     
- Misses        281      435     +154     
- Partials       45       69      +24     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.