Improve documentation of ErrorTracker.Error kind field (#154)

We thought about it and came to the conclusion that it would be good to
change the type declaration for the module. Open for discussion, because
that could potentially be a breaking change of the library if users
previously had implementations of `ErrorTracker.Ignorer` that matched on
non-string types (although, arguably those implementations would never
have worked).

This addresses #153.

---------

Co-authored-by: Manmohan Krishna <manmohan.krishna@maersk.com>

Author: tjarratt

guides/Getting Started.md

@@ -191,6 +191,19 @@ The `ErrorTracker.Ignorer` behaviour allows you to ignore errors based on their
 
 When an error is ignored, its occurrences are not tracked at all. This is useful for expected errors that you don't want to store in your database.
 
+For example, if you had an integration with an unreliable third-party system that was frequently timing out, you could ignore those errors like so:
+
+```elixir
+defmodule MyApp.ErrorIgnores do
+  @behaviour ErrorTracker.Ignorer
+
+  @impl ErrorTracker.Ignorer
+  def ignore?(%{kind: "Elixir.UnreliableThirdParty.Error", reason: ":timeout"} = _error, _context) do
+    true
+  end
+end
+```
+
 ### Muting Errors
 
 Sometimes you may want to keep tracking error occurrences but avoid receiving notifications about them. For these cases,

lib/error_tracker/schemas/error.ex

@@ -12,7 +12,16 @@ defmodule ErrorTracker.Error do
 
   use Ecto.Schema
 
-  @type t :: %__MODULE__{}
+  @type t :: %__MODULE__{
+          kind: String.t(),
+          reason: String.t(),
+          source_line: String.t(),
+          source_function: String.t(),
+          status: :resolved | :unresolved,
+          fingerprint: String.t(),
+          last_occurrence_at: DateTime.t(),
+          muted: boolean()
+        }
 
   schema "error_tracker_errors" do
     field :kind, :string