to answer both of your question, yes i tried the card withtout changing settings, hot spot was 107 degrees and 88 degrees in apex at around 165fps.
You can try undervolting and it is helpful not only to lower temps but also to improve performance. Modern GPU's are very temperature sensitive and anything you can do to keep hot spot temps below 90-95C only serves to improve performance.
Arranging fans to blow cool air into the intake area of your GPU's coolers, or even removing a case side cover, can help a lot with temperature. This may be the reason you're seeing high temps (107C) and instability, especially if you have a glass-side case with a vertical mount GPU.
I'm not sure what voltages are attainable for a 6700XT, but my 6800XT likes to have a setting of around 1068-1081mV for gaming. It can handle one or two passes of TimeSpy benchmark as low as 1050mV at a fairly heavy overclock but it will start throwing a lot of math errors in GPU Folding@Home unless kept above 1081mV even at stock clocks.
But be a bit cautious when increasing the power limiter. While it will most likely handle being pushed up to the max allowed, if you do the GPU will hit hot-spot temperatures of around 90C sooner and then start pulling back on clocks. Test it with back-to-back TimeSpy runs to see when it's hurting and not helping. Maxing the power limiter is usually beneficial only if you have extremely good cooling for the GPU...something like liquid cooling...but people do it a lot and think they've done a trick since it might give a terrific score with one pass at TimeSpy.
Also, AMD has imposed a hard power limit on Navi2 GPU's in the BIOS which mean they they can not exceed certain settings. So if the board partner has already increased the power limit with their (sanctioned) OC model then you only get the remainder to the AMD imposed hard limit. That may explain what you're seeing. There are ways around this with
RedBiosEditor (RBE) if you're really interested, but it's definitely hardcore.
And lastly, undervolting may seem a bit strange if you're also monitoring the GPU voltage when in-game where it might hit voltages well above what you set it at. The way it's been explained to me is it's more of an off-set adjustment for introducing a bias to the boost algorithm to make it not hit as high of a voltage in each P-state as it would have before. So it will still hit higher voltages but not likely to hit the max, 1150mV in my case with stock settings