Optimise Bind-Pop-PushB bytecode sequences

[rmhsilva]

In many cases, the following bytecode is generated by the naïve compiler:

....
 |   10 | BIND     x
 |   11 | POP
 |   12 | PUSHB    x
...

The last two instructions cancel each other out, and can be deleted entirely. This is a great simple optimisation to add!

Background reading: https://teal-book.condense9.com/vm/source-to-success-compilation.html#compile

Brief implementation thoughts:

• in compiler.py...
• create a function called optimise_bytecode for this and future optimisations
• create this optimisation in remove_redundant_bytecode or something
• call optimise_bytecode from CompileTopLevel.make_function