compiler:remove unneccessary clone in final code gen

[Timosdev99]

this pr unnecessary cloning of intermediate byte code blocks and per instruction cloning during final code gen. the compiler now moves byte code block by value and processes instruction by reference

• move intermediate_bytecode.bytecode into code_blocks instead of cloning Vec<IntermediateInstruction>
• refactor compile_block and codegen_jump to use burrowed IntermediateInstruction/IntermediateValue rather than clone each instruction

the previous implementation duplicate the entire intermediate program and cloned every instruction inside a hot loop, causing unnecessary heap allocation

[TomWambsgans]

Hi, thanks for the PR. Since the compilation performance is not currently an issue (and overall it's never a bottleneck since we need need to compile once, and then it does not impact the proving performance of the snarks), I would suggest we keep the compilation pipeline as simple as possible (even though it's already too much complex..), in particular Cloning is not an issue, and I think it's slightly simpler than std::mem::take(&mut intermediate_bytecode.bytecode). Also, after formatting, your PR does not reduce the nb of LoC, so I would suggest not to merge it for now. Does this reasoning make sense to you?