another ping: OP: Could you please check the maximal error?

... and of course, the test that fails

Hmm. I'm afraid it's still failing for me.

I checked out the most recent octave-image repo and installed the "image-2.13.0+" library (from changeset 1382:af78d1f1ec4c, which I think is the commit you're referencing here) and re-ran the tests. Still one failure for imadjust:

>> pkg load image
>> runtests2 -pkg image
[...]

  /Users/janke/Library/Application Support/Octave.app/4.4.1/pkg/image-2.13.0+/imadjust.m  PASS     35/36
                                                                FAIL      1
[...]
Summary:

  GNU Octave Version: 4.4.1 (hg id: 1f46d371968c + patches)
  Tests run on angharad.local (macOS) at 29-May-2020 03:45:20
  Test execution time: 00:00:48

  PASS                             1697
  FAIL                                1
  XFAIL (expected failure)           34

  Failed tests:
     /Users/janke/Library/Application Support/Octave.app/4.4.1/pkg/image-2.13.0+/imadjust.m

Anything else I can do to help diagnose?

I have pushed a small fix which adds tolerance of eps to the
failing test
(http://hg.code.sf.net/p/octave/image/rev/af78d1f1ec4c)
please check this

I'm getting a couple test failures in the image package on macOS 10.14.6.

Under Octave 4.4.1:

Summary:

  GNU Octave Version: 4.4.1 (hg id: 1f46d371968c + patches)
  Tests run on angharad.local (macOS) at 14-Feb-2020 23:43:00
  Test execution time: 00:01:13

  PASS                             1694
  FAIL                                4
  XFAIL (expected failure)           34

  Failed tests:
     /Users/janke/Library/Application Support/Octave.app/4.4.1/pkg/image-2.12.0/imadjust.m
     /Users/janke/Library/Application Support/Octave.app/4.4.1/pkg/image-2.12.0/immse.m
     /Users/janke/Library/Application Support/Octave.app/4.4.1/pkg/image-2.12.0/mean2.m

Under Octave 5.2.0:

  GNU Octave Version: 5.2.0 (hg id: eb46a9f47164)
  Tests run on angharad.local (macOS) at 14-Feb-2020 23:39:09
  Test execution time: 00:01:05

  PASS                             1697
  FAIL                                1
  XFAIL (expected failure)           34

  Failed tests:
     /Users/janke/octave/image-2.12.0/imadjust.m


XFAIL items are known bugs or expected failures.

Please help improve Octave by contributing fixes for them.
octave:13> test imadjust
***** test
 img = rand (4, 4, 2, 3, 2);
 adj = zeros (4, 4, 2, 3, 2);
 in  = reshape ([ 3  5  7  9 11 13 15 17 19 21 23 25;
                 97 95 93 91 89 87 85 83 81 79 77 75] / 100, [2 2 3 2]);
 out = reshape ([ 5  7  9 11 14 15 17 19 21 23 25 27;
                 95 93 91 89 87 85 83 81 79 77 75 73] / 100, [2 2 3 2]);
 gamma = reshape (0.6:.1:1.7, [1 2 3 2]);
 for p = 1:2
   for q = 1:3
     for r = 1:2
       adj(:,:,p,q,r) = imadjust (img(:,:,p,q,r), in(:,p,q,r),
                                  out(:,p,q,r), gamma(1,p,q,r));
     endfor
   endfor
 endfor
 assert (imadjust (img, in, out, gamma), adj)
!!!!! test failed
ASSERT errors for:  assert (imadjust (img, in, out, gamma),adj)

  Location  |  Observed  |  Expected  |  Reason
(3,1,2,2,1)    0.34036      0.34036      Abs err 5.5511e-17 exceeds tol 0 by 6e-17
(2,2,2,2,1)    0.38731      0.38731      Abs err 5.5511e-17 exceeds tol 0 by 6e-17
(4,2,2,2,1)    0.15268      0.15268      Abs err 2.7756e-17 exceeds tol 0 by 3e-17
(2,3,2,2,1)    0.22302      0.22302      Abs err 2.7756e-17 exceeds tol 0 by 3e-17
(4,4,2,2,1)    0.50159      0.50159      Abs err 1.1102e-16 exceeds tol 0 by 1e-16

That last one looks like it's maybe just a tolerance error where it's testing for an exact instead of approximate result.