Allow settings for keep alives and idle connection timeouts for scrape configs

[mattdurham]

Allow the usage of custom keep alive duration for scrape configs (both normal and scraping service). This hopefully can be done via the custom DialerFunc

[mattdurham]

Ideally should be able to pass to the instance.go:newScrapeManager

[rfratto]

The custom DialerFunc wouldn't do it, but we can use config.WithKeepAlivesDisabled and config.WithIdleConnTimeout from prometheus/common/config.

[jnyi]

Is there any progress on this one, we are seeing 404 not found issue with kubernetes API for kubelet proxy: https://stackoverflow.com/questions/49275380/node-proxy-404-page-not-found, if we could disable keep alives for the connection and idle conn timeout, hopefully this could be mitigated.