[Refactor] Refactor to use new fileio API in MMEngine.

[mzr1996]

Remove all FileClient and use the new API in MMEngine.

[codecov]

Codecov Report

Base: 0.02% // Head: 91.11% // Increases project coverage by +91.09% 🎉

Coverage data is based on head (7b31971) compared to base (b8b31e9).
Patch has no changes to coverable lines.

❗ Current head 7b31971 differs from pull request most recent head 3b8ed07. Consider uploading reports for the commit 3b8ed07 to get more accurate results

Additional details and impacted files

@@             Coverage Diff              @@
##           dev-1.x    #1176       +/-   ##
============================================
+ Coverage     0.02%   91.11%   +91.09%     
============================================
  Files          121      138       +17     
  Lines         8217    10221     +2004     
  Branches      1368     1611      +243     
============================================
+ Hits             2     9313     +9311     
+ Misses        8215      687     -7528     
- Partials         0      221      +221     

Flag	Coverage Δ
unittests	91.11% <ø> (+91.09%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
mmcls/apis/inference.py	0.00% <0.00%> (ø)
mmcls/datasets/transforms/compose.py
mmcls/models/heads/efficientformer_head.py	93.10% <0.00%> (ø)
mmcls/models/backbones/mobilevit.py	91.15% <0.00%> (ø)
mmcls/models/backbones/mvit.py	92.46% <0.00%> (ø)
mmcls/models/retrievers/__init__.py	100.00% <0.00%> (ø)
mmcls/models/backbones/swin_transformer_v2.py	89.47% <0.00%> (ø)
mmcls/utils/progress.py	66.66% <0.00%> (ø)
mmcls/models/necks/reduction.py	100.00% <0.00%> (ø)
mmcls/models/retrievers/base.py	100.00% <0.00%> (ø)
... and 128 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[yingfhu]

Is it necessary to use backend.join_path instead of join_path since join_path already uses enable_singleton=True.

The time saving might be marginal and is only for load annotations at the beginning, which is not affecting training data time.

use join_path seems more unified. Otherwise comments must be added for those use backend.join_path for better understanding

[yingfhu]

LGTM