Use MapSets for performance

Not noticeable with current database size, but lays a good foundation
for larger libraries.

For example, given a list of 300 artist IDs, checking if an ID is in the
list takes between 30 and 40 ms (for an ID towards the end of the list),
while the MapSet version takes consistently between 15 and 20 ms.

lib/music_library/collection.ex

@@ -56,7 +56,7 @@ defmodule MusicLibrary.Collection do
         where: fragment("records.purchased_at IS NOT NULL"),
         select: r.value
 
-    Repo.all(q)
+    q |> Repo.all() |> MapSet.new()
   end
 
   defp base_search do

lib/music_library/records.ex

@@ -103,7 +103,7 @@ defmodule MusicLibrary.Records do
   def get_all_artist_ids do
     q = from ar in ArtistRecord, distinct: true, select: ar.musicbrainz_id
 
-    Repo.all(q)
+    q |> Repo.all() |> MapSet.new()
   end
 
   def get_artist_records(musicbrainz_id) do

lib/music_library/wishlist.ex

@@ -27,7 +27,7 @@ defmodule MusicLibrary.Wishlist do
         where: fragment("records.purchased_at IS NULL"),
         select: r.value
 
-    Repo.all(q)
+    q |> Repo.all() |> MapSet.new()
   end
 
   defp base_search do