Modify default histogram boundaries to align with the limit sizes

[Aoao54]

What changed?

Modify default histogram boundaries to align with the limit sizes.

Why?

For now, the default bucket sizes of some metrics did not align with the default limit sizes. Such as the default threshold of history_size_error is 50MB but the default max bucket sizes is 16MB. We can see the error rate in the prometheus and alert on it without any config change.

How did you test it?

Tested locally.

Potential risks

Is hotfix candidate?