Remove SM_75 architecture when compiling with CUDA 10.0

[MoisesHer]

Description

Avoid compilation of code for SM_75 architecture when compiling with CUDA 10.0 to solve bug #17713
GPUs based on SM_75 architecture are able to run code generated for SM_70 architecture:
https://docs.nvidia.com/cuda/turing-compatibility-guide/index.html#turing-volta-compatibility

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

• Changes are complete (i.e. I finished coding on this PR)
• Code is well-documented
• To the best of my knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

• Remove SM_75 as target architecture in linux_cu100 compilation

[mxnet-bot]

Hey @MoisesHer , 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: [unix-cpu, edge, website, centos-gpu, windows-gpu, sanity, miscellaneous, clang, centos-cpu, windows-cpu, unix-gpu]

---

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.

[MoisesHer]

@mxnet-bot run ci [miscellaneous, unix-cpu]

[mxnet-bot]

Jenkins CI successfully triggered : [miscellaneous, unix-cpu]

[leezu]

@MoisesHer: @mseth10 ran into a size limitation problem while fixing the CD (and switching to cmake) and disabled SM_75 on all staticbuilds for now (#18205)