Implement more retries on timeout errors

[drh-stanford]

Currently, GeoMonitor retries twice on Faraday connection errors, like Faraday::Timeout. We may want to bump up the retries and backoff, and/or the timeout options.

See

geo_monitor/lib/geo_monitor/requests/wms.rb

Lines 41 to 47 in e4cfce3

	conn = Faraday.new(url: url)
	begin
	conn.get do |request|
	request.params = request_params
	request.options.timeout = 10
	request.options.open_timeout = 10
	end

and

http://www.rubydoc.info/github/lostisland/faraday/Faraday/Request/Retry