Support input of non three-channel image

[MGAMZ]

Motivation

Current Implementation does not support images with non 3-channels

Modification

Add in_chans augment and pass it to PatchEmbed layer.

[codecov]

Codecov Report

Attention: 1074 lines in your changes are missing coverage. Please review.

Comparison is base (896964b) 65.16% compared to head (f378f36) 62.77%.
Report is 17 commits behind head on dev.

Files	Patch %	Lines
mmpretrain/models/multimodal/ram/bert.py	0.00%	449 Missing ⚠️
mmpretrain/models/multimodal/ram/ram.py	0.00%	144 Missing ⚠️
mmpretrain/models/multimodal/clip/clip.py	0.00%	139 Missing ⚠️
mmpretrain/models/multimodal/ram/gradio_demo.py	0.00%	54 Missing ⚠️
mmpretrain/models/multimodal/llava/modules.py	0.00%	44 Missing ⚠️
mmpretrain/models/multimodal/ram/openset_utils.py	0.00%	42 Missing ⚠️
...retrain/models/multimodal/clip/clip_transformer.py	0.00%	41 Missing ⚠️
mmpretrain/models/multimodal/ram/utils.py	0.00%	38 Missing ⚠️
mmpretrain/models/multimodal/minigpt4/minigpt4.py	0.00%	28 Missing ⚠️
mmpretrain/datasets/minigpt4_dataset.py	0.00%	27 Missing ⚠️
... and 12 more

Additional details and impacted files

@@            Coverage Diff             @@
##              dev    #1866      +/-   ##
==========================================
- Coverage   65.16%   62.77%   -2.39%     
==========================================
  Files         362      375      +13     
  Lines       26182    27177     +995     
  Branches     4165     4309     +144     
==========================================
  Hits        17061    17061              
- Misses       8501     9495     +994     
- Partials      620      621       +1     

Flag	Coverage Δ
unittests	62.77% <0.27%> (-2.39%)	⬇️

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

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

[MGAMZ]

Github Shows only 1 file changed, but codecov shows many files have been changed which is not the truth.