RequestTimeout

It seems that the recent changes to request_timeout don't allow enough time for all the devices on larger systems and/or with slower computers to respond before being timed out. See discussion here https://github.com/MindrustUK/Heatmiser-for-home-assistant/issues/69. As a short term-fix suggest you increase default value to 15 as described. If this isn't acceptable long term due to performance issues we could have an initial (or periodic) higher value to count the devices and then set the request time out accordingly.