dslx_reference improvements

[gittywithexcitement]

I'm going to record some things I noticed that need improvement:

• [x] In the Precedence table, mis-rendered || and |
  • https://github.com/google/xls/pull/1292
• [x] In the Precedence table, >>> should not be there because it does not exist
  • https://github.com/google/xls/pull/1292
• [x] concatenate operator is missing from Precedence. I believe it's the same precedence as + and - (confirmed below)
  • https://github.com/google/xls/pull/1298
• [ ] .. the range expression is missing from Precedence. I believe it's lower than everything else (lower than boolean or)
• [ ] / and % are in the table but not mentioned in binary expressions. Both operators are worth describing their precise semantics (e.g. is % the remainder or the modulo function?)
• [ ] as is never introduced as a binary expression or operator. It looks like this section is about as. But I think the current way things are organized and described could be improved. E.g., as is clearly a binary expression as it takes two arguments. Why not make it a subsection under Binary Expressions? This suggestion also applies comparison operators and concatenation operator. Anything that's in the precedence table should probably be grouped as a unary or binary expression/operator.
• [x] (...), the grouping expression, is never is never mentioned as an expression, nor is its precedence mentioned (I assume it's the highest precedence?)
  • https://github.com/google/xls/pull/1298
• [ ] true and false are never introduced as a bit type
• [ ] 'expression statements' are never documented. E.g. Foo();
• [ ] let expressions are not expressions. They are statements. In the past they were expressions but they aren't now. Update the wording and move to the statement section.

[gittywithexcitement]

Proof that concatenation is the same precedence as + - and also left associative.

  assert_eq(u4:0b1 ++ u1:0b1 +  u5:0b1, u5:0b100);
  assert_eq(u4:0b1 +  u4:0b1 ++ u1:0b1, u5:0b101);

  assert_eq(u4:0b1 -  u4:0b1 ++ u1:0b1, u5:0b1);
  assert_eq(u4:0b1 ++ u1:0b1 -  u5:0b1, u5:0b10);

[gittywithexcitement]

Proof that let is not an expression:

0036:   assert_eq(let x=u32:6; x, u32:6);
~~~~~~~~~~~~~~~~~~^-^ ParseError: Expected start of an expression; got: keyword:let