Oh, nifty. How did you do that?
Long ago, I wrote a user script to augment the FR posting form. It fixes some formatting deficiencies and adds a UI for injecting commonly needed HTML. Nothing fancy.
When the current bug surfaced, and after waiting for immediate fix which didn't come, I added some code that scans input about to be posted and substitutes HTML entities for any characters outside the 7-bit range. The server lets entities pass. It mangles the corresponding UTF-8, however, which is the current bug.
The result is that I can post arbitrary Unicode and have it display correctly. The downside is, any attempt to quote my posts results in garbage, absent the appropriate client-side assist!