[ISSUE-1168] Read whitespace in tokenstream and reconstruct source content

[sshailabh]

• Read whitespace to a separate channle in tokenstream.
• Added Template.getSourceText() for exact source reconstruction with whitespace preservation. Enables formatters, linters, and IDE tooling.

Implementation:

• Whitespace tokens sent to channel 1 (not skipped) in HbsLexer.g4
• BaseTemplate stores tokenStream + token span (start/end indices)
• TemplateBuilder.visitTemplate() and visitBody() set token spans during parsing
• getSourceText() uses TokenStream.getText(Interval) for exact reconstruction
• ForwardingTemplate delegates getSourceText() to wrapped template

Example:

handlebars.compileInline("{{ foo }}").text();          // "{{foo}}"   (normalized, Existing)
handlebars.compileInline("{{ foo }}").getSourceText(); // "{{ foo }}" (exact, New)