[R4R] fix: memory leak issue with diff protocol

[j75689]

Description

broadcastDiffLayers is not properly closed, the memory accumulates too many resources and causes OOM.

goroutine 8845520 [select, 3426 minutes]:
github.com/ethereum/go-ethereum/eth/protocols/diff.(*Peer).broadcastDiffLayers(0xc081628d20)
	github.com/ethereum/go-ethereum/eth/protocols/diff/peer.go:47 +0xab
created by github.com/ethereum/go-ethereum/eth/protocols/diff.NewPeer
	github.com/ethereum/go-ethereum/eth/protocols/diff/peer.go:41 +0x2f0

goroutine 7119763 [select, 3531 minutes]:
github.com/ethereum/go-ethereum/eth/protocols/diff.(*Peer).broadcastDiffLayers(0xc10bcd4e40)
	github.com/ethereum/go-ethereum/eth/protocols/diff/peer.go:47 +0xab
created by github.com/ethereum/go-ethereum/eth/protocols/diff.NewPeer
	github.com/ethereum/go-ethereum/eth/protocols/diff/peer.go:41 +0x2f0

goroutine 11443031 [select, 3272 minutes]:
github.com/ethereum/go-ethereum/eth/protocols/diff.(*Peer).broadcastDiffLayers(0xc0f9bb3260)
	github.com/ethereum/go-ethereum/eth/protocols/diff/peer.go:47 +0xab
created by github.com/ethereum/go-ethereum/eth/protocols/diff.NewPeer
	github.com/ethereum/go-ethereum/eth/protocols/diff/peer.go:41 +0x2f0

Rationale

broadcastDiffLayers will be started in NewPeer; need to call peer.Close when RunPeer fails. otherwise, this goroutine will not be closed.

Example

panic.log

Changes

Notable changes:

• close the peer on failure in RunPeer function.

[brilliant-lx]

LGTM

[forcodedancing]

LGTM