fix bug of gaussian_target, update unittest of heatmap

[tianyuandu]

1. Fixed the bug in gen_gaussian_target while multiple gts in one class but only got one peak in the heat map.
2. Expand to 3 gt_bboxes in test_corner_head_encode_and_decode_heatmap and two of them are the same class.
3. Replace expand with repeat in corner_head.py to avoid shallow-copy cases, which may generate unexpected results in 2.

codebase	batchsize	mAP
Current	16 x 6	41.2
Current	10 x 5	still running
Last benchmark	16 x 6	41.1
Last benchmark	10 x 5	41.3
Paper	10 x 5	40.5

[codecov]

Codecov Report

Merging #3543 into master will decrease coverage by 0.06%.
The diff coverage is 94.11%.

@@            Coverage Diff             @@
##           master    #3543      +/-   ##
==========================================
- Coverage   60.92%   60.85%   -0.07%     
==========================================
  Files         205      205              
  Lines       13802    13809       +7     
  Branches     2340     2341       +1     
==========================================
- Hits         8409     8404       -5     
- Misses       4966     4971       +5     
- Partials      427      434       +7     

Flag	Coverage Δ
#unittests	60.85% <94.11%> (-0.07%)	⬇️

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

Impacted Files	Coverage Δ
mmdet/models/dense_heads/corner_head.py	74.31% <93.75%> (+0.22%)	⬆️
mmdet/models/utils/gaussian_target.py	100.00% <100.00%> (ø)
mmdet/models/roi_heads/test_mixins.py	45.16% <0.00%> (-8.61%)	⬇️
mmdet/models/dense_heads/rpn_test_mixin.py	77.41% <0.00%> (-6.46%)	⬇️
mmdet/core/bbox/samplers/sampling_result.py	73.01% <0.00%> (-1.59%)	⬇️

---

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 49b3f37...d64efd4. Read the comment docs.

[hellock]

ping @tianyuandu

[tianyuandu]

codebase	batchsize	mAP
Current	16 x 6	41.2
Current	10 x 5	still running
Last benchmark	16 x 6	41.1
Last benchmark	10 x 5	41.3
Paper	10 x 5	40.5

[hellock]

@tianyuandu The linting failed.