enhancement: allow more missing upstream info in parse_nginx_log

[ex5]

There's a valid use case when upstream_response_length, upstream_response_time, upstream_status are sometimes unavailable but the rest of the log message still matches the upstreaminfo format (e.g. static files served by the same virtual host). Here's an example of such a message:

1.1.1.1 - - [09/Oct/2023:10:49:43 +0200] "GET /media/cache/f4/b4/f4b494deadbeefc8.jpg HTTP/1.1" 200 32235 "https://example.com/page/" "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/117.0.0.0 Safari/537.36" 951 0.000 [upstream-production-service-1234] [upstream-production-service-1234] - - - - 7d1f3deadbeefcbb3bc

Currently, this kind of message generates a parsing error and some logging data is lost. It would be useful if instead the rest of the available data was still parsed correctly.

[pront]

Hi @ex5, this change seems reasonable to me, thanks!

Two things we still need:

• Add a changelog.md entry for this change
• Update the Vector docs here with a new example with arguments like the ingress_nginx_upstreaminfo_valid_missing_upstream.
  • cue fmt path/to/file and make check-component-docs are the commands you will need.

[pront]

Looks good, thank you.