And I'm not really sure why you're saying support is at 23k because this is what I'm looking at... 25k or so was tested twice during August and Feb, before finally breaking out at March. Now it seems like it has flipped back to being resistance.

It's because $23K is the volume support, whereas there isn't a lot of support volume between $24K and $26.5K. It's also very close to the 200 Day MA currently around $22.5K that within the coming days will be at $23K. As you can see there is increasing volume support from $25K down to $23K, but similar to the correction in March, price came down to the peak of this volume support before reversing to the upside.

In summary, the same reason I was bullish about $20K acting as support, while others were fearing much worse prices, is more of the less the exact same reason why I see $23K acting as a support level. Arguably there is more chance of $23K acting as support than there was of $20K back in March, as back then the 200 DMA was barely rising, whereas now it's been in an uptrend for the past 3 months.
It's also partially due to the market mentally. Most were/are expecting $25K to act as support, somewhat similar to those who thought $22K/$23K would act as support back in March, whereas instead it failed. If most people expect old resistance to turn into new support it probably won't, especially with price effectively in a neutral zone between the 50 Day MA (declining bearish) and 200 Day MA (rising bullish).
Naturally I'd have a completely different opinion if for example the re-test of $25K was in line with some other support level, like the 50 Day MA, or otherwise the 200 Day MA, but this isn't the case. Old resistance on it's own turning into new support just seems unlikely to me, especially now the 200 WMA has broken. Whereas the confluence of support as described at $23K looks a lot more appealing to me.