Random test generation is flawed; they keep generating wrong results for huge tests and even with reduced board size (like 5x5) they keep generating wrong expected values, albeit more rarely:

XBZBK
FHBAP
XBEJH
TFLJO
EBQYG
expected:<2> but was:<3>

It's a worthwhile Kata, but the description needs to be clearer. Maybe talk about bimodal/multimodal and clarify that we're looking for how many modes the data set contains (not the value of the mode or how many times it occurs).

oh my! points for cleverness but not readibility

Amazing solution! I was hoping to write something like this but couldn't. Well done :)

Whoa, you've already prepared it for complier

You are creating map on each iteration

Thank you.

Very clever solution !!!

I've never known about class RegExp in Kotlin. I used to user class Regular only. How to import class RegExp in Kotlin?

The description is still inadequate; you did not fix all problems and you introduced new problems with what you did fix.

Never mind. It's retired.

For future kata you want to publish, please, get help for your descriptions. There's a Gitter channel for authoring kata; you can ask for help there. Looking at this and earlier kata of yours, writing descriptions is really not your strong suit.

I removed the note and modified the prolog. Please see again the desciption.

Description has issues, see below. These issues are still present.

Please do not close issues unless you have outside confirmation the problems have been rectified.

Doing nothing and closing the issue is frowned upon.

Ok.

The string's length at least 2.

Then you contradict yourself:

Note: When the string length is not enough or empty string, you just give the distance is 0.

The description also has numerous missing words and grammatical and other errors. Somebody, not necessarily you if your English is not good enough, should give it a good go-over.

Examples should come after specifications. No new specifications should be introduced after the examples have started.