Ah, you meant a non-Roman text. That possibility occurred just after I’d posted. I can confirm your problem with the (I’m using OS 10.14.6 in case that is pertinent, and my locale is ‘en-gb’).
Checking further, I can use the capitalised version of the text with
.lowercase' and get the lowercase example string. I can do the reverse with the lower-case example and .uppercase
Well, this is a workaround for now. For me, with a note holding one of the two examples in the $Text of two different notes, this agent matches both notes:
$Text.lowercase.icontains("γάιος ιούλιος καίσαρας")
Note that this query doesn’t affect the source $text, but instead tests an all-lowercase version of the source.
Here’s my test file: Greek-test.tbx (87.9 KB)
It may work as a stop-gap. I’ve only tested the examples given but the file should allow you to test a bit more.