Autopkgtests failing with liboctave4

[Łukasz Zemczak]

With the upgrade of octave to 4.2.1 (and the sobump to liboctave4) the matlab2tikz reliable autopkgtests started failing. Test log follows:

```matlab
suite = @ACID;
alltests = [1:105];
reliable = [2:5 7 9 11:16 19:20 22:24 28 32 35:37 39:40 42:44 46:57 59:62 64:77 79:85 87:90 93 97:100];
unreliable = [1 6 8 10 17:18 21 25:27 29:31 33:34 38 41 45 58 63 78 86 91:92 94:96 101:105];
failReliable = [2:5 7 9 11 13:16 19:20 22:24 28 32 35:37 39:40 50 52:53 55:57 60:61 64:70 73:74 77 79 81 83:85 87:90 93 97 99:100];
passUnreliable = [];
skipped = [12 42:44 46:49 54 59 62 71:72 75:76 80 82 98];
```

| | Pass | Fail | Skip | Total |
| :--------- | ---: | ---: | ---: | ----: |
| Unreliable | 0 | 32 | 0 | 32 |
| Reliable | 1 | 54 | 18 | 73 |
| Total | 1 | 86 | 18 | 105 |

The main reason for this to happen is that with the new octave, the matlab2tikz test output now has additional 'formatting' included in the axis definition, e.g.

 every outer x axis line/.append style={black},
 every x tick label/.append style={font=\color{black}},
 every x tick/.append style={black},

This blocks octave from getting promoted to -updates.

[Łukasz Zemczak (sil2100)]

After looking closer at the results of the tests one way to fix this would be to actually just change the test hash values to the new ones as these are the only changes that have appeared. They look like harmless as the output is identical to the previous ones.

[Łukasz Zemczak (sil2100)]

By 'the output is identical' I mean how the pdf generated out of the tex file.

[Łukasz Zemczak (sil2100)]

For reference, logs to a failure against the new octave:
https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-artful/artful/amd64/m/matlab2tikz/20170612_093659_5f751@/log.gz

[Łukasz Zemczak (sil2100)]

Forwarded proposed hash change to upstream with the following PR:
https://github.com/matlab2tikz/matlab2tikz/pull/995

I have tested a local build of the package in a PPA and ran it through autopkgtests and there are no reliable failures anymore. So this essentially fixes the issue - I might wait for upstream to comment first though before uploading.

[Brian Murray (brian-murray)]

The autopkgtests are passing on mantic now so I'm setting this bug task to Fix Released.