Use unambiguous float duration values

[antkmsft]

Same rationale as #516 and #527: for example, when representing 35.621 as float64, 35.62100000000000221689333557151257991790771484375 and 35.621 are the same number (kind of how 42 and (int)42.1 are the "same number" too), first being explicit and accurate, second one is what will get you to the first one when rounding, assuming 64 bits. (and for 32 bits, 35.621 would be represented as 35.620998382568359375). I.e. can't reliably expect 35.621.ToString() to be 35.621 and not 35.62100000000000221689333557151257991790771484375 on all implementations, and should implementation give us "35.62100000000000221689333557151257991790771484375", it won't be incorrect, but cadl-ranch would fail.

If we use 35.625, it is not ambiguous: 35.625 can be exactly represented as IEEE754 floating point, there is no room for misinterpretations, there is exactly one string representation for it, not two.
So, in this PR, we're updating 35.621 to 35.625, and 46.781 to 46.75 for the same reason.

[changeset-bot]

🦋 Changeset detected

Latest commit: 086b650

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 2 packages

Name	Type
@azure-tools/cadl-ranch-specs	Patch
@azure-tools/cadl-ranch	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR