Fix possible IMA in amp_multicast fusion

[mk-61]

Description

The patch fixes possible CUDA Invalid Memory Access when fusing amp_multicast op with inputs of different sizes.

It has been observed on Yolo3 model from GluonCV.

Checklist

Essentials

• Changes are complete (i.e. I finished coding on this PR)

@ptrendx

[mxnet-bot]

Hey @mk-61 , Thanks for submitting the PR
All tests are already queued to run once. If tests fail, you can trigger one or more tests again with the following commands:

• To trigger all jobs: @mxnet-bot run ci [all]
• To trigger specific jobs: @mxnet-bot run ci [job1, job2]

---

CI supported jobs: [windows-cpu, unix-gpu, miscellaneous, edge, unix-cpu, centos-gpu, centos-cpu, sanity, website, windows-gpu, clang]

---

Note:
Only following 3 categories can trigger CI :PR Author, MXNet Committer, Jenkins Admin.
All CI tests must pass before the PR can be merged.

[ptrendx]

LGTM