Enhance stability through redundancy of providers

[longfin]

Currently, the bridge relies only one ethereum node at some points. so when that endpoint becomes unavailable, bridge couldn't be recovered without restarting.¹

To alleviate this phenomenon, it would be helpful to achieve the redundancy of providers and configure it in a classic failover style. (See also)

Footnotes

1. Of course, smoothing this out with issues like the Expose the endpoint for health checking #210 is also a good approach. ↩