Fix dtype issues

[SkafteNicki]

Before submitting

• Was this discussed/approved via a Github issue? (no need for typos and docs improvements)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure to update the docs?
• Did you write any new necessary tests?

What does this PR do?

Fixes #492
Fixes #484

Reverts some of the changes in #462.

• Removes the dtype property. Metrics state can have different dtypes (like any other nn.Module) so the property probably does not make sense for all metrics
• Fixes bug with mixed precision training: half, float, double will now have no effect on the dtype of metric states to prevent that metrics gets auto-cast during mix-precision. Instead if the user really wants to change the dtype of all metric states they can now do it using the set_dtype method.

PR review

Anyone in the community is free to review the PR once the tests have passed.
If we didn't discuss your PR in Github issues there's a high chance it will not be merged.

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #493 (3921a24) into master (f96c717) will decrease coverage by 23%.
The diff coverage is 10%.

@@           Coverage Diff           @@
##           master   #493     +/-   ##
=======================================
- Coverage      95%    72%    -23%     
=======================================
  Files         132    132             
  Lines        4681   4676      -5     
=======================================
- Hits         4462   3365   -1097     
- Misses        219   1311   +1092     

[Borda]

@ethanwharris mind have look? ^^

[tchaton]

LGTM !

[ethanwharris]

Awesome, confirmed that this fixes my issue 😃 thanks @SkafteNicki !

[Borda]

@SkafteNicki seems some tests are running for too long with latest versions