Improve shard error message formatting

[kertal]

We introduced a modal providing details to shard errors in v7.4.0 with PR #41649

It allows users to get details about shard errors, before just the message "x of y shards failed" was detailed.

Here's how the modal looks like

The following should be improved

• Row expansion

  • Avoid using bold in error messages
  • Avoid the big gaps between labels and values
  • Reduce the font size in the header
  • Row expansion should just display a list of field values, rather than trying to construct a meaningful sentence.

• Table

  • Move node to row expansion to provide more space for the other columns (it's not very useful).

[elasticmachine]

Pinging @elastic/kibana-data-discovery (Team:DataDiscovery)

[sophiec20]

Note that this can look particularly scary in very large environments and esp with long (CCS) index names.

[kertal]

Thx @sophiec20 for providing additional feedback, I do agree we should aim to fix this dated UI. Apart from the layout, I think the information presented here to debug shard errors, something like this would be useful for CCS failures, telling users about timeouts in certain CCS clusters or something like this (Once this information would be available in a search request result, which isn't currently the case)

[sophiec20]

Yes, and we can make it better in increments, starting with a more succinct, cleaner and simpler presentation of today's elasticsearch errors. As a follow up, we can work with es to improve error handling.

[andreadelrio]

Given that values of these columns can get quite long (see index and node names in Sophie's examples) why don’t we try an inverted layout instead?

This new layout does away with the unnecessary repetition of information we have in the current one. See:

Some things to consider:

• When there's only one page, I recommend we hide the pagination component. That would make the modal cleaner.