Fix E203 false positive for slices in format strings

[sciyoshi]

Summary

The code later in this file that checks for slices relies on the stack of brackets to determine the position. I'm not sure why format strings were being excluded from this, but the tests still pass with these match guards removed.

Closes #10278

Test Plan

Still needs a test. Test case added for this example.

[github-actions]

ruff-ecosystem results

Linter (stable)

✅ ecosystem check detected no linter changes.

Linter (preview)

ℹ️ ecosystem check detected linter changes. (+0 -2 violations, +0 -0 fixes in 1 projects; 42 projects unchanged)

RasaHQ/rasa (+0 -2 violations, +0 -0 fixes)

ruff check --no-cache --exit-zero --ignore RUF9 --output-format concise --preview

- rasa/nlu/utils/bilou_utils.py:292:58: E203 [*] Whitespace before ':'
- rasa/nlu/utils/bilou_utils.py:294:49: E203 [*] Whitespace before ':'

Changes by rule (1 rules affected)

code	total	+ violation	- violation	+ fix	- fix
E203	2	0	2	0	0

[charliermarsh]

I added this in #8654, but I'm not sure why.

[charliermarsh]

I probably copied it from missing_whitespace, but again, I don't know why.

[charliermarsh]

I think this is correct, though @dhruvmanila can check my work.

[dhruvmanila]

I think this is correct as well, thanks.

The reason the logic exists in missing_whitespace is to avoid adding whitespace after : in cases where there's a format specifier in a f-string because it then changes the semantic meaning (f"{data:.3f}" vs f"{data: .3f}").