fix(coverage): proper instruction for 1st branch anchor

[grandizzy]

Motivation

Re https://t.me/foundry_support/55383
PR #8414 fixed the branch coverage report (issue reported in #7784 ) by properly mapping if/else source bodies

foundry/crates/evm/coverage/src/analysis.rs

Lines 236 to 241 in 5af9d16

	match node.attribute::<Node>("falseBody") {
	// Both if/else statements.
	Some(false_body) => {
	// Add the coverage item for branch 1 (false body).
	// The relevant source range for the false branch is the `else` statement
	// itself and the false body of the else statement.

but it also made the wrong assumption that the first branch opcode after JUMPI should be pc + 1

foundry/crates/evm/coverage/src/anchors.rs

Lines 149 to 153 in 5af9d16

	ItemAnchor {
	item_id,
	// The first branch is the opcode directly after JUMPI
	instruction: pc + 1,
	},

This is not true because the evaluation is done for the opcode before JUMPI (so JUMPI is pc + 1 and next opcode is pc + 2)

foundry/crates/evm/coverage/src/anchors.rs

Lines 134 to 136 in 5af9d16

	// Check if we are in the source range we are interested in, and if the next opcode
	// is a JUMPI
	if is_in_source_range(element, loc) && bytecode[pc + 1] == opcode::JUMPI {

Solution

Proper source mappings fixes #7784 and #4294 (require and assert branches properly reported now), hence this PR reverts the instruction for first branch and adds back branching for asserts
Added unit tests for all require/assert branches
#4309 #4310 and #4315 tests were passing with wrong branch opcode, will follow up with a fix for them

[mattsse]

cool