[Lens] Misleading percentages shown in field stats when the field is only on some docs

[wylieconlon]

The main reason it's misleading is the footer which says for each field that the field is in "99% of 5,076 documents", which has ambiguous meaning. Will a user interpret this setting as:

• We analyzed 99% of documents, and here are the values we foun
• The field is available on 99% of documents
• Confusion?

This is very closely related to not using the sampler aggregation.

Potential solutions

My preferred solution is to use this display to indicate that the field is available on X% of documents, without sampling.

An alternative option is to indicate the sampling separately from the "existing" percent, effectively introducing a "Missing" value to each field.

[elasticmachine]

Pinging @elastic/kibana-app (Team:KibanaApp)