Fixes #39. Spaces and source-language name for language attribute

[TobiasZawada]

The asciidoctor manual accepts spaces at the beginning of the language attribute of code blocks. Furthermore, the language attribute can also be a named attribute source-language.

So

[source,emacs-lisp]
----
	(message "Hello world")
----

is the same as

[source, emacs-lisp]
----
	(message "Hello world")
----

and as

[source, source-language = emacs-lisp]
----
	(message "Hello world")
----

Only the first version is fontified as emacs-lisp in origin/master at commit b0702bdeb8d799720661766936c36ec3b454c609.

[bbatsov]

I'd add here some comments with the snippets from the manual, so it's clear what we're matching against. Ideally we should also have some matching unit tests, but this can be done later.

[TobiasZawada]

Thanks for the review.
The asciidoctor manual is not clear about the 2nd code block attribute. So I had a look at the Ruby code. There I misinterpreted the document attribute source-language as block attribute. I corrected this with the last commit.

[TobiasZawada]

It looks like the source-style is applicable to any fenced block.
At the end of the following page they use it with the open block (which is very general): https://docs.asciidoctor.org/asciidoc/latest/blocks/open-blocks/

So maybe, we should take a more general approach and just test the style of each block.
Maybe, even asciidoc-Formatters still work in source blocks whose style is not listing or literal.

We can think about this later.
The changes in this branch/pull-request are nevertheless an improvement.

[TobiasZawada]

The rabbit-hole is quite deep: asciidoctor/asciidoctor-pdf#180

[TobiasZawada]

@bbatsov ... and even deeper ...

:source-highlighter: pygments
:myBlock: source,ruby,subs="verbatim,quotes"

[{myBlock}]
----
class SomeClass
  for i in 1..5 do
      "Hello *world*".print
  end
----

:saveOldBlock: {myBlock}
:myBlock: source,python,subs="verbatim,quotes"

[{myBlock}]
----
class AnotherClass
  for i in 1..5 do
      "Hello *world*".print
  end
----

:myBlock: {saveOldBlock}
[{myBlock}]
----
class BehindTheLocalDef
  for i in 1..5 do
      "Hello *world*".print
  end
----

Attribute substitutions are carried out in attribute lists before parsing: https://docs.asciidoctor.org/asciidoc/latest/attributes/positional-and-named-attributes/#attribute-list-parsing

[TobiasZawada]

@bbatsov As we have seen, the rabbit hole is quite deep.
Let us accept these changes for now that at least those source blocks are highlighted that we have discovered in the spec so far, i.e., listings, open blocks and literal blocks.