fix onnx conversion of maxunpool2d

[grimoire]

Hi
This PR fix ONNX conversion of maxunpool2d by adding offset to indices.

[codecov]

Codecov Report

Merging #243 (1e5ad4c) into master (2827d93) will decrease coverage by 0.64%.
The diff coverage is 58.44%.

❗ Current head 1e5ad4c differs from pull request most recent head 5f4f2ec. Consider uploading reports for the commit 5f4f2ec to get more accurate results

@@            Coverage Diff             @@
##           master     #243      +/-   ##
==========================================
- Coverage   81.79%   81.14%   -0.65%     
==========================================
  Files         152      153       +1     
  Lines        7223     7436     +213     
  Branches     1066     1088      +22     
==========================================
+ Hits         5908     6034     +126     
- Misses       1191     1271      +80     
- Partials      124      131       +7     

Flag	Coverage Δ
unittests	81.14% <58.44%> (-0.65%)	⬇️

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

Impacted Files	Coverage Δ
...ckbones/encoder_decoders/decoders/plain_decoder.py	66.29% <0.00%> (-24.62%)	⬇️
mmedit/models/restorers/liif.py	65.07% <65.07%> (ø)
mmedit/datasets/pipelines/utils.py	94.64% <100.00%> (+5.35%)	⬆️
mmedit/models/restorers/__init__.py	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2827d93...5f4f2ec. Read the comment docs.

[RunningLeon]

We may add the reasons why we do this change with comments here.