Release variable in windows data conversation

[directionless]

Release variable in cimDatetimeToUnixtime

i think I got all the returns. It passes CI. I can download and run the resultant osqueryd. Not sure if there's much else to manually test here

Fixes: #7018

[zwass]

Can we use the scope_guard strategy for this? As seen in https://github.com/osquery/osquery/pull/7024/files#diff-572b7f350484fd1cc8be8e82ef3948646c04e248e249ef6c0a2a1b7dda592b06R103-R104.

That helps us ensure that the Release() is called even if more code paths are added later.

[directionless]

Can we use the scope_guard strategy for this?

That sounds much cleaner. I was wishing for a go style defer, let's see if CI says I got the syntax right

[zwass]

Yes scope guard is awesome! Very similar to Go's defer, though cool that it works on a scope basis with RAII.

[zwass]

Docs (https://docs.microsoft.com/en-us/windows/win32/api/combaseapi/nf-combaseapi-cocreateinstance) seem to indicate we should be good with this.

Upon failure, *ppv contains NULL

[directionless]

I downloaded the CI artifact and ran it on windows, select * from processes seemed to work.

[directionless]

CI artifact still seems okay for select * from processes which I think his this code path.

Kinda makes me wondering about sprinkling this scope_guard release pattern everywhere