add some big improvements to parse error messages

change the parser to not attempt to parse the elements following
'from' unless there is a actual 'from'
improve the symbol parser to try to deal with issues when symbols are
  next to eachother with no intervening whitespaces
improve number literal parsing to fail if there are trailing letters
  or digits which aren't part of the number and aren't separated with
  whitespace
add some code to start analysing the quality of parse error messages

TODO

@@ -1,4 +1,5 @@
 continue 2003 review and tests
+docs: how to run the tests
 touch up the expr hack as best as can
 left factor as much as possible (see below on notes)
 table expression in syntax:
@@ -14,11 +15,41 @@ look at fixing the expression parsing completely
 represent natural and using/on in the syntax more close to the
    concrete syntax - don't combine in the ast
 
+review the token parsers, and make sure they have trailing delimiters
+   or consume bad trailing characters and fail (e.g. 1e2e3 in a select
+   list parses as '1e2 e3' i.e. '1e2 as e3'
+   split the general symbol and operator parsing, and make it tighter
+   in terms of when the symbol or operator ends (don't allow to end
+   early)
+approach: review the lexical syntax, create complete list of
+   tokens/token generators. Divide into tokens which must be followed
+   by some particular other token or at least one whitespace, and ones
+   which can be immediately followed by another token. Then fix the
+   lexing parsers to work this way
+whitespace/comments
+integers
+numbers
+string literals
+keywords
+operator symbols <>=+=^%/*!|~&
+non operator symbols ()?,;"'
+identifiers
+quoted identifiers
+
+identifiers and keywords are ok for now
+there are issues with integers, numbers, operators and non operator
+   symbols
+
+
+review places in the parse which should allow only a fixed set of
+   identifiers (e.g. in interval literals)
+
 decide whether to represent numeric literals better, instead of a
    single string - break up into parts, or parse to a Decimal or
    something
 
 rough SQL 2003 todo, including tests to write:
+can multipart identifiers have whitespace around the '.'?
 multipart string literals
 national, unicode, hex, bit string literals, escapes
 string literal character sets
@@ -92,7 +123,7 @@ additional stuff review:
 interval stuff
 collate clause
 aggregate functions: lots of missing bits
-
+complete list of keywords/reserved keywords
 
 
 review areas where this parser is too permissive, e.g. value