Holding an unordered set is surprisingly inefficient for large sets.
Switching to just a vector makes e.g. deserialization of large modules
30% faster. We pay in terms of memory usage though, so if there is ever
a use case where we are storing the same code objects multiple times, we
might want do add a deduplication algorithm which cleans up the vector
every now and then.

R=thibaudm@chromium.org

Bug: v8:11164
Change-Id: I3983ee7f6f04ea7678b8da49fb5cec369693dbc3
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2647260
Commit-Queue: Clemens Backes <clemensb@chromium.org>
Reviewed-by: Thibaud Michaud <thibaudm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#72325}