Off by one(1) error in column addressing: fixing originalPositionFor() API for indexed maps + added test adapted from the 0.6.x dev tree

[GerHobbelt]

The main work was done in SHA-1: 88efe61

• fixing originalPositionFor() API for indexed maps + added test adapted from the 0.6.x dev tree: it turned out that due an off-by-1 error in the column value, half of the mapping nodes weren't hit by the originalPositionFor() internal search code, while it should find those.

The added tests are augmented with extra logging for diagnosis: run those tests with & without the given patch to observe the bug & change.

---

Before this pullreq, the 0.7.x/0.8.x did not test this situation; I ran into trouble with indexed maps crashing and acting crazy. The added test ensures there's no regression.

WARNING / NOTE: this pullreq includes #394 and #398 as required prerequisites: without those, this won't fly, as the crashes/failures in the test set will simply move to other spots in the code.
Hence this is more complex issue as #394 and #398 must be fixed before this one can work.

See the commit log for this one: patch-3 = #394 and patch-7 = #398 which are visible below as git merge commits.

[coveralls]

Pull Request Test Coverage Report for Build 561

• 7 of 8 (87.5%) changed or added relevant lines in 3 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.1%) to 88.28%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
lib/source-map-consumer.js	3	4	75.0%

Totals
Change from base Build 550:	-0.1%
Covered Lines:	825
Relevant Lines:	913

---

💛 - Coveralls

[GerHobbelt]

Ha! This is a very close simile of #382: I browsed the pullreq list just now and discovered I had ripped that one's unit test into my 0.6.x and then went for 0.8.x and redid the work.
Turns out my edit is -1 decrement of the rightmost term, while #382 @Motix88 had +1 increment for the leftmost term, making these fixes work the same.

However 😉 , the -1 approach is already present a little lower in the code in this little chunk that sits just below the place where we both patched the code:

    return section.consumer.originalPositionFor({
      line: needle.generatedLine - (section.generatedOffset.generatedLine - 1),
// ^^^^^^^^^^^^^^^^^^^^ note this existing line ^^^^^^^^^^
      column:
        needle.generatedColumn -
        (section.generatedOffset.generatedLine === needle.generatedLine
          ? section.generatedOffset.generatedColumn - 1
          : 0),
      bias: aArgs.bias
    });

Also, this pullreq makes all tests pass and for that I had to mix in the other pullreqs mentioned earlier. Without them, the test kept failing; see also the console diagnostics in the commits.

[bomsy]

Closing as fixed by #507