I've worked around this issue - this is just to note it in case anyone else finds themselves in the same position.
Last weekend I found web pages taking at least 4 seconds, sometimes longer to load - and it looked like DNS queries had randomly started to have significant delays. Using nslookup it was clear this was the problem; a new query would time out, but it would then work if re-queried (albeit unreliably). There was no configuration change on my end that resulted in this.
The system was using https-dns-proxy forwarding to the standard Quad9 servers. Switching to Google's DNS servers resolved the issue. I seemed to be able to also use the Quad9 servers directly (i.e. from PC configuration) - although I didn't rigorously test that. My short term solution was to use the Google servers, and I have now moved to dnscrypt-proxy2 which appears to be working well with Quad9 servers.
When trying to diagnose the problem it was not clear at all what might be interfering with the queries, and Quad9 were not reporting problems. It could be something strange happening at my end or with my ISP but I struggle to think what might result in this behaviour. Having worked on it for a while I've done the work-around and moved on, but if this evolves into a more common issue and/or wants to suggest further diagnostics I'd be happy to try and recreate it and look more closely.
