Solr has a limit on the number of facets - which has nothing to do with the Query.MaxItems btw (that is for Sitecore queries). You should be able to set facet.limit to -1 to get all results.
Not sure if in the current implementation by Sitecore there is another solution possible already, but this was given ...
You could change the default field mapping in your index configuration (i would create a custom indexconfiguration based on the default):
and remove the suffix from the format (this example is for text fields, you should change it for ID-types):
I came across this issue when upgrading a solution to 9.3. Querying the database for "illegal" characters did not find any. Disabling media indexing also did not help and is not an option for this solution either.
The root cause of the issue could be some encoding mismatch, f. ex. some characters that has been pasted in by editors. SolrNet does not ...
I ended up opening a support ticket with Sitecore and they identified this as a known bug.
The hotfix I received today is SC Hotfix 418336-1 SXA 184.108.40.20689.
Consider contacting them for the latest hotfix, as it may include other fixes to SXA.