Although a second ISP link can perhaps avoid an issue with one ISP taking down your Internet connectivity, there can be much more you might do for HA. For example, HA often avoids using only one router unless it's a higher end model that provides hardware redundancy in the "box", and even then, the chassis itself can be a single point of failure. Even with a router that provides hardware redundancy, or multiple routers, depending on the level of HA you're attempting to achieve, a single site can be a single point of failure, and that's not limited to your site. For example, when using two ISPs how do the links physically run? Could a single back-hoe cut both links? Do both links terminate at the same POP? And so on.
Regarding load sharing, most routing protocols will support static ECMP (equal cost multiple path) load sharing. Which means, if the router has multiple equal cost paths to the same destination, it will round robin (egress) flows across the multiple paths. However, the router does not normally take into account actual link load or issues with a link. For example, a link could be saturated or encountering a high level of corruption of frames/packets on the link, and the router would normally just keep directing flows to that link.
The one Cisco technology that I'm aware of that can perform dynamic load balancing and/or move traffic off links having issues is Cisco's PfR (Performance Routing). This technology is available on many of Cisco's routers, although I believe it may often require a license upgrade to use it.
PfR can do a wonderful job of balancing egress, and although it has some support for ingress balancing, that's takes much more to support. Often, though, if you control both sides of a network, the other side's egress may balance out the far side's ingress. However, when using general Internet (working with hosts on the Internet, not under your control), you often have few to zero options for ingress load balancing.
