fix(l2): return None when branch becomes empty in remove

[sashass1315]

In BranchNode::remove() return None when the branch has zero children and no value after removal, instead of returning Mutated. Node::remove() relies on new_root.is_none() to mark the subtree as empty, and Trie::remove() uses that signal to reset the root to the empty trie (keccak(RLP_NULL)). LeafNode::remove() and ExtensionNode::remove() already adhere to this contract, so this change aligns BranchNode with the established semantics and prevents persisting a structurally empty branch at the root which would otherwise break the empty trie invariant.

[mpaulucci]

is this a bug? A perf optimization? How did you get into this? Why l2?

[Copilot]

Pull Request Overview

This PR fixes a bug in BranchNode::remove() where an empty branch (no children and no value) would return Mutated instead of None. The fix ensures that when a branch becomes completely empty after removal, it properly signals this by returning None, which allows Node::remove() to propagate the empty state and Trie::remove() to reset the root to the empty trie hash.

Key changes:

• Added an early return in BranchNode::remove() to check if the branch is empty after removal
• Returns (None, value) when both children and value are empty, aligning with LeafNode and ExtensionNode behavior

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.