Fix crash when creating thread

[timothyqiu]

Fixes #53046

There are two different crashes:

1. Did not check MethodBind * for null: it crashes when the target method does not exist.
2. Did not validate target instance: if the target instance is freed before the new thread is started, it's undefined behavior to call its methods.
   • ObjectDB::instance_validate() is gone on master, so I changed target_instance from plain old Object * to ObjectID in order to validate it.

[akien-mga]

Needs a rebase after #53053.

[RandomShaper]

Just for the records, it's worth noting that, by doing the check inside _start_func() instead of start(), this is also protecting against a race condition where the original thread (or some other) frees the object just after having created the new thread, before the latter has had a chance to obtain the object pointer. That part of the protection will only take us so far, but I don't think it's a bad thing either. The important fact is that the important timespan covered by this protection, the same that we would get by having the check in start(), is still there, only extended to account for a potential race condition.

[akien-mga]

Thanks!