This will cause automatic retries to be skipped, I'm not sure we want to do this.
Any thoughts?

This ConnectTimeoutException is subclass of InterruptedIOException which is in exceptionBlacklist of RetryHandler. So retryRequest() will cancel retry for this exception currently. Which one do you prefer? To retry on this exception or return an error to caller?

btw, the SocketTimeoutException is similar with ConnectTimeoutException. It's also a subclass of InterruptedIOException. The retryRequest() catch SocketTimeoutException and return an error to caller so I think maybe you want to do the same for ConnectTimeoutException.

I see two issues with your merge request.
First, you're trying to catch error, which happens only if you use the library incorrectly by catching the issue afterwards. I'd rather see checking the url on set or before the request is created. Library shouldn't really automatically sanitise your generated URLs.
Second, you're fixing two different issues in your merge request, it's quite ambiguous.

I'll incorporate the ConnectTimeoutException myself.