Improve Expression Parsing

[jalextowle]

Description

The current strategy for parsing expressions is bulky, unmaintainable, and most likely incorrect. A better strategy is needed.

Strategy

1. Once it is determined that an expression needs to be parsed, tokens should be collected until a semicolon is found (the terminator of an expression statement).
2. The collection of tokens should be parsed using a function:

fn parse_expression(expression: &mut Vec<lex::Token>) -> ParseNode

3. The parse_expression function should likely use a LR parsing strategy, or at the very least employ the Shunting-yard Algorithm.