GeezerVison

• if we received an invalid HELLO (or probably also READY) event, DSharpPlus would trigger the resume procedure. if the shard never successfully connected, this is fine, because the resume URL and session ID will be null and it'll trigger a hard reconnect, but if it did successfully connect before, it would reuse that resume URL and session ID for a resume attempt that may now be invalid. this would cause shards to stall infinitely in an attempt to resume before a session was established in the first place

• if discord closed the gateway without an error code, DSharpPlus would crash the shard. we will now attempt to resume.

• if discord sent an invalid gateway payload, DSharpPlus would crash the shard to restart from a clean slate. it now waits for five consecutive invalid events before rendering judgement so as to allow for transient errors

• MultiShardOrchestrator would previously always immediately reconnect a failed shard without considering that many shards can fail at once. this adds a concurrency limit (currently of 1, since the actual gateway concurrency limit is more perfidious and 1 should guarantee success) to multi-sharded reconnects

• This pull request did not involve AI in any way

• All features in this pull request were tested.

this PR initially only intended to fix the first issue, but the list has grown