That’s incorrect. The mutation load model isn’t about traits at all. That’s been a misdirection from the start. The model is about mutation accumulation and the effect of the load on fitness. Nothing more or less.
That’s because it’s a redefinition of fitness that obfuscates the problem of mutation load.
That’s not true at all. The fitness decline has indeed been observed in reality, but in most cases the timescale needed to witness this is too long with respect to the time we have been observing it.
One example I have seen where a species has been acknowledged to have gone extinct from mutation load would be the wooly mammoth.