That "short-term improvement in mental health cannot be extrapolated to longer term outcomes" is the part that needs data if you want to overturn the findings about it being protective against suicide.

What findings would they be?

You posted two links, one which said nothing at all either for against so-called "gender-affirming care" (GAC) reducing suicide rates. The other one was a poor-quality study with a high drop-out rate and of a completely different demographic to the majority of teens receiving GAC today. It showed an increase in depression, and no time trend, both of which suggest that their major finding of reduced suicide was just a statistical artefact, not a real effect. If it were a real effect, it should show a time trend and a decrease in depression.

The whole field is, sadly, in a very poor state:

• Most studies on GAC are poor quality, at high risk of bias, and with low confidence in their results including landmark studies such as De Vries et al. (2011).
• The evidence base for GAC is very poor with few studies using controls (whether matched or unmatched) and lack of randomisation. With these studies involving so many subjective elements and being at such high risk of bias, the lack of blinding is a severe problem.
• GAC is also at high-risk of confounding due to the placebo effect.

Finland, Sweden and France have moved away from GAC in the last few years, expressing serious doubts that the benefits are greater than the risk of side-effects and harms. Norway has recently done the same. And after the Tavistock scandal, the British NIH have distanced themselves from from GAC for youth as well.

The so-called "Dutch Protocol" is intentionally conservative towards patients suffering from confounding mental illnesses. Whereas in the US especially, having multiple mental illnesses is treated as all the more reason to fast-track "gender affirming care" and rush children and teens into irreversible treatments.