Fix_1368 Sigma 18-35mm f/1.8 DC HSM (0.27 only)

[lbschenkel]

See #1368. On Canon bodies and lens firmware 1.x, this lens identifies itself using
model number 150. Starting with firmware 2.x, this lens now identifies
itself using model number 368.

Exiv2 was correctly identifying the former but not the latter.

This PR also fixes a pre-existing bug in which exiv2 was not
correctly disambiguating lenses using model number 368.

[clanmills]

This looks good. Thanks very much.

My only comment is to point out that can run exiv2 (or any command) more than once in a test. Commands is an array, as are retval, stderr, stdout. If you'd like to change it, I will merge later. If you want to leave it "as is", I'll merge.

I will cut'n'paste my instructions to you about the test harness in to the book. So, while helping you, the book has also developed. I'll mention you on page 2.

If you'd like to do more work on Exiv2, I would be delighted to have your help. In August, I recruited a young Chinese engineer @LeoHsiao1 who has done marvellous work on rewriting bash_tests in python. So Leo meet Leonardo. Leonardo meet Leo.

I've changed the title to match my naming style.

[clanmills]

Another comment about this. I haven't really studied the issue here. If it's straight-forward to port this PR to master can you change the trailer of the title (0.27) to be (0.27->master) to be sure that the release engineer (probably me) deals with it.

If you intend to submit a different PR for 'master', please change the trailer to (0.27 only).

[lbschenkel]

My only comment is to point out that can run exiv2 (or any command) more than once in a test. Commands is an array, as are retval, stderr, stdout. If you'd like to change it, I will merge later. If you want to leave it "as is", I'll merge.

No problem, I'll improve it.

If you'd like to do more work on Exiv2, I would be delighted to have your help.

I'm afraid that I won't be able to commit to that. I have a newborn at home and a number of other projects that I'm the maintainer for. :-) But I think I will be able to submit future fixes unassisted, if I ever encounter a similar issue.

If you intend to submit a different PR for 'master', please change the trailer to (0.27 only).

Master does not have this particular bug, because in master this is the only lens with model 368. But I can submit the new tests if you like.

Pull request has been modified.

[lbschenkel]

Improved test case to avoid running multiple commands.

[clanmills]

Good Job. Thanks very much. Please submit a PR to master with an update to the test suite.