External patches should have checksums

[sitsofe]

While this initially sounds like Issue #15631 I think there's a case saying that there should be basic checks against corrupt (accidental or otherwise) external patches.

An example of this could be a linked patch that is on a publicly modifiable webpage. It seems reasonable to assert the patch is likely in the same state as when the formula was accepted...

An extension of this could be a --insecure option that when used once no longer requires patch checksums / allows usage of resources that can't be checksumed thus preserving a degree of backwards compatibility.

[MikeMcQuaid]

Duplicate of #11956 and #21648.