If you do not reuse passwords and one of them does leak, then the only thing affected is the site/service that was compromised. Hence the word "unique" in the title.
Leaking a password on your side is an unknown unknown, so password rotation is not a bad practice on its own for a security conscious person: It limits a leak in time.
Mandatory password rotation is a whole different kettle of fish, as it pushes users to lower password quality. Infosec policy was required to balance 2 conflicting needs, and the past has thougt us we balanced wrong.
Even if you rotated your password frequently, there would still be a large window of compromise. Password rotation only helps with very strange attack scenarios, and passwords themselves aren't really good enough for anything where security actually matters.
I would personally push away from passwords on the whole at this point. SSO is probably more secure for most users. Plenty of websites only support username+password auth, and given how bad most passwords are... I might even go so far as to suggest that username+TOTP is instantly more secure than that, especially with proper rate limiting as you should have anyways. (Yes, I know TOTP is "supposed" to only be a second factor.)
WebAuthn takes this to the next level and promises a future where you can use a strong single factor to log in, without any opportunity for phishing or credential compromise... but most implementations I've seen still require a fallback password mechanism. There are understandable reasons for this right now, but it is unfortunate.
There original threat model for forced password rotation was supposedly based on hash cracking time. This is a stupid threat model; the guy from NIST who wrote it back in the 80s admitted it was based on no research but was added arbitrarily because it sounded good at the time.