fix EMAModel.use_ema_warmup

[chenguolin]

What does this PR do?

I do not understand for a long time why use_ema_warmup is forced to be True in EMAModel.__init__(). It is really inconvenient for really proferming fixed-decay EMA.

And it is also unreasonable to still use warmup by cur_decay_value = (1 + step) / (10 + step) when use_ema_warmup=False.

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline?
• Did you read our philosophy doc (important for complex PRs)?
• Was this discussed/approved via a GitHub issue or the forum? Please add a link to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests?

Who can review?

General functionalities: @sayakpaul @yiyixuxu @DN6

[sayakpaul]

I think this becomes a breaking change for folks using this class. It's better to deprecate the bevahiour.

[chenguolin]

Maybe adding a deprecating logging? The current decay behavior is unreasonable for me:

1. we can't set self. use_ema_warmup in the EMAModel initialization, which means input arg use_ema_warmup is useless;
2. EMA decay is still warmup when self.use_ema_warmup=False.

[sayakpaul]

Deprecation is the way to go. We cannot introduce a breaking change like this.