Initial support for OpenACC syntax

[jefflarkin]

This PR creates initial support for OpenACC directives, based on the existing OpenMP syntax.

Changes:
Made OpenMP sentinel less greedy
Added initial support for OpenACC directives

Future work:

• I would love to implement the actual grammar so that clauses that are invalid on particular directives do not match. Currently this does the same simple matching as the OpenMP syntax on which I based this change.
• Does not currently match the arguments to data clauses, may be a nice addition, but they are a bit complex to match
• Does not currently match optional clause modifiers, like readonly: and zero:

[gnikit]

The one thing that would be useful would be to add a Fortran file where you test these directives. I added some myself under test/resources I found it extremely helpful while making changes to visually inspect that I have not broken anything and it will be useful in the future when we figure out how we can unittest the syntax highlighting.

Also, as I said in my tracking issue, could you think of any cases where the OpenMP functionality would fail with the change that you made?

[jefflarkin]

Sure. I will add the test cases I was using when I'm back at that computer on Monday.

Regarding the OpenMP matching, what I found was that with the existing matching it would sometimes erroneously match an OpenACC directive with the OpenMP syntax. For instance 'acc parallel do' is not valid, but it would get matched from the OpenMP. I made the matching less greedy and as best as I could tell still match correctly.

[jefflarkin]

@gnikit I added resources/test/openacc_support.f90. I'm working through multi-line matching at the moment. This doesn't appear to work as expected in either the OpenMP or OpenACC. If either & is removed, it matches properly, but if both are there it doesn't match.

I included a test at the bottom to test that the OpenMP is not over-matching the OpenACC, but is still correctly matching on its own.

[jefflarkin]

I submitted #225 related to line continuations, but since this is true for the existing OpenMP and OpenACC, I don't think it should block the merging of this issue. The fix should be the same for each.

[jefflarkin]

Fixes #225

[jefflarkin]

@gnikit I believe this is ready to merge and put out in the market. Do you agree?

[gnikit]

Awesome job @jefflarkin. Thanks for taking the time to do this.
I had a quick play around with my OpenMP and the added OpenACC test and all looks good.

It truly is a shame that we cannot unify the openmp.json and openacc.json into one file since they have a lot of duplicate code, but I think it is better to add support now for the openacc and improve on it in the future, if need be.

I left some minor comments, but no major changes.

I look forward to seeing this in master.

[gnikit]

I don't know if @krvajal you want to have a look. All is good. I will add the updated CHANGELOG in a bit.

[jefflarkin]

Fixes #184