Use UInt32 for stored character indices

This greatly reduces the size of the TaggedRange data structure on 64
bit systems, at the cost of not being able to parse files larger than
4GiB. But that seems like a reasonable tradeoff, for now. (We can
parameterize later, if really necessary)

It apperas that this one change makes parsing ~10% faster.

Author: c42f

src/parse_stream.jl

@@ -79,8 +79,8 @@ Information about preceding whitespace is added for use by the parser.
 """
 struct SyntaxToken
     kind::Kind
-    first_byte::Int
-    last_byte::Int
+    first_byte::UInt32
+    last_byte::UInt32
     # Flags for leading whitespace
     is_dotted::Bool
     is_suffixed::Bool
@@ -122,17 +122,17 @@ TODO: Optimize this data structure?  It's very large at the moment.
 struct TaggedRange
     head::SyntaxHead # Kind,flags
     orig_kind::Kind  # Kind of the original token for leaf tokens, or K"Nothing"
-    first_byte::Int  # First byte in the input text
-    last_byte::Int   # Last byte in the input text
-    start_mark::Int  # Index of first emitted range which this range covers
+    first_byte::UInt32  # First byte in the input text
+    last_byte::UInt32   # Last byte in the input text
+    start_mark::UInt32  # Index of first emitted range which this range covers
 end
 
 head(range::TaggedRange)       = range.head
 kind(range::TaggedRange)       = kind(range.head)
 flags(range::TaggedRange)      = flags(range.head)
 first_byte(range::TaggedRange) = range.first_byte
 last_byte(range::TaggedRange)  = range.last_byte
-span(range::TaggedRange)       = last_byte(range) - first_byte(range) + 1
+span(range::TaggedRange)       = 1 + last_byte(range) - first_byte(range)
 
 #-------------------------------------------------------------------------------
 struct ParseStreamPosition