Add option to ignore parens in lisp/scheme block comments

[mnacamura]

Closes #66
(This PL is built on top of #79, so please merge #79, if it is okay, before this.)

Scheme and Common Lisp support nested block comment:

#| depth 1 comment started
    #| depth 2 comment |#
    here is still in comment of depth 1
|#

This PL enables parinfer-rust to ignore parens inside the nested block comment.

For doing so, I did a bit of refactoring for function on_char() and now it dispatches depending not only on char but also on parser context (i.e., reading code, reading comment, reading string, reading Janet long string, etc.) It works fine for me practically but introduces a failing test case:

test indent_mode ... FAILED
test smart_mode ... ok

failures:

---- indent_mode stdout ----
thread 'indent_mode' panicked at 'assertion failed: `(left == right)`
  left: `"(let [a 1] ;\n  ret)"`,
 right: `String("(let [a 1]) ;\n  ret")`: case 538: text', tests/cases.rs:34:9
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace.


failures:
    indent_mode

test result: FAILED. 8 passed; 1 failed; 0 ignored; 0 measured; 0 filtered out

Looks like related to JSON serialization. @eraserhd Could you kindly give me any suggestion to fix this if you have time and if interested in this PL?Fixed!

[mnacamura]

Another glitch has been found. Block comment in one line works:

(let ([x 1])
  #|comment|#
  x)

But in multi lines, it fails:

(let ([x 1
  #|com
  ment|#
  x)

[mnacamura]

Fixed.

[eraserhd]

Thanks for this!

Please add an entry to CHANGELOG.adoc and add your name to "Contributors" in README.adoc if you like.

[mnacamura]

I pushed (hopefully) the final bug fix for this PL and updated the documents. I feel really honored to be counted as a contributor. Thanks!

[eraserhd]

Thank you for the work!