[FIX] Parsing messages with multiple markdown matches ignore some tokens

[c0dzilla]

@RocketChat/core

Closes #9854 #9758

The message body was containing a random string of the form '=!=randomInteger=!=' many times whenever the message had multiple markdown matches(for example a url with inline code). This was because the tokens were being unmounted in the order in which they were attached during parsing, leading to misses when a token was overwritten by another token as in the case of mutiple matches. This pr changes unmounting order to LIFO to fix this.

Example:

now parsed as

[rodrigok]

Thanks @c0dzilla

What do you think about use reverse to keep the code cleaner?

example:

- for (const {token, text} of message.tokens) {
+ for (const {token, text} of message.tokens.reverse()) {

[c0dzilla]

@rodrigok I've made the change.

[DeanVanGreunen]

Thanks for fixing Guys 🥇