log: increase precision in log's timestamp to nanoseconds

[lecaros]

Adds nanoseconds precision to the logs Fluent Bit writes.

It's possible in some cases that the underlying implementation of the time functions won't provide nanoseconds precision. In those cases, the precision will be to microseconds, which is still better than to the second. (see here)

Addresses #9983

---

Fluent Bit is licensed under Apache 2.0, by submitting this pull request I understand that this code will be released under the terms of that license.

[lecaros]

ping @edsiper

[lecaros]

ping @edsiper @pwhelan
This could be considered a breaking change. We're on time to add it to 4.0

[pwhelan]

I feel v4.0 is a reasonable time to make it the default, the risk should be minor. @edsiper How do you feel about it?

[edsiper]

@lecaros @pwhelan wondering if this implies a higher CPU usage, is there a way you can provide some perf changes ?

[lecaros]

Thanks @edsiper. Done.
And thanks @leonardo-albertovich for the reviews

[leonardo-albertovich]

Just one more comment because this popped in my notifications. Are we going to fix the inconsistency in flb_time_get or not?

I think it should be done in this PR in order to ensure that we have the same behavior in all platforms, otherwise the format string which prints seconds.nanoseconds (which might already be misleading) will show seconds.microseconds in macos but the user wouldn't be aware of it.

If you need help with the patch let me know @lecaros.

[edsiper]

@leonardo-albertovich I recommend triaging that potential issue in a separate PR (so both can make the cut for 4.0.3)

[leonardo-albertovich]

I think that'd be quite cumbersome because it means this PR which depends on that would have to wait and get rebased.

Is there a specific reason to do so?

[edsiper]

moving to next milestone due to code conflict and resolution

[lecaros]

code conflict resolved. @edsiper