Parse SUBSTRING FROM syntax in all dialects, reflect change in the AST

[lovasoa]

always accept both substring function syntaxes in all dialects,
and ensure SUBSTRING expression are always formatted with the same syntax as the one they were parsed with.

This both simplifies the code (making it easier to maintain as databases evolve) AND creates a higher-fidelity syntax tree for dialects that support both syntaxes

fixes #1168

[coveralls]

Pull Request Test Coverage Report for Build 8266220365

Details

• 19 of 20 (95.0%) changed or added relevant lines in 2 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.01%) to 87.87%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/parser/mod.rs	14	15	93.33%

Totals
Change from base Build 8239291082:	0.01%
Covered Lines:	20486
Relevant Lines:	23314

---

💛 - Coveralls

[lovasoa]

@alamb : I think this is the best solution on the long term. This creates a higher-fidelity AST, and avoids having to maintain an up to date supports_substring_from_for_expr for all databases, which leads to issues like #1168

[alamb]

I agree @lovasoa -- this design is consistent with the rest of sqlparser-rs and the design of "parse the syntax and leave semantics to the user"

Since it is now possible to distinguish between the two syntactic forms via AST I believe users can implement whatever behavior they desire.

THank you very much

[alamb]

Thanks again @lovasoa