Add `error_type` granularity to pills displaying run state

We've got a bunch of different types of errors that we track in Lightning and we use them to create different visuals using the step_icon function:

case {reason, error_type} do
  {nil, _any} -> [:pending, "text-gray-400"]
  {"success", _any} -> [:success, "text-green-500"]
  {"fail", _any} -> [:fail, "text-red-500"]
  {"crash", _any} -> [:crash, "text-orange-800"]
  {"cancel", _any} -> [:cancel, "text-grey-600"]
  {"kill", "SecurityError"} -> [:shield, "text-yellow-800"]
  {"kill", "ImportError"} -> [:shield, "text-yellow-800"]
  {"kill", "TimeoutError"} -> [:circle_ex, "text-yellow-800"]
  {"kill", "OOMError"} -> [:circle_ex, "text-yellow-800"]
  {"exception", ""} -> [:triangle_ex, "text-black-800"]
  {"lost", _nil} -> [:triangle_ex, "text-black-800"]
end

In this list, you can see the "exit_reason" on the left of the tuple, and the "ErrorType" on the right. Note that this is for a step, not a run. We do not yet provide any error_type granularity for run state pills.

It's very important for our users to see that one run was killed for TimeoutError while another was killed for running out of memory OOMError. Right now, they'll only see "Killed". This issue is to split up the run state pills to introduce error_type granularity.

There is another related issue (#1350) which allows the user to filter the history page based on runs with particular states and with particular error_types.