fix broken quartiles method in isoutlier

[AvanishSalunke]

1. quartiles method in isoutlier always crashed, because dim was being passed into the wrong argument of quantile. While fixing it, also found the center value formula was wrong - it used the median instead of the average of the 25th and 75th percentiles, which is what MATLAB actually returns.
2. Also removed a docstring line that incorrectly said ThresholdFactor isn't supported for quartiles.

BEFORE:

octave:2> A = [57 59 60 100 59 58 57 58 300 61 62 60 62 58 57];
octave:3> [TF, L, U, C] = isoutlier(A, 'quartiles')
error: Q(3): out of bound 2 (dimensions are 1x2)
error: called from
    isoutlier>quartiles_method at line 382 column 3
    isoutlier at line 341 column 8

AFTER:

octave:2> A = [57 59 60 100 59 58 57 58 300 61 62 60 62 58 57];
octave:3> [TF, L, U, C] = isoutlier(A, 'quartiles')
TF =

  0  0  0  1  0  0  0  0  1  0  0  0  0  0  0

L = 52.375
U = 67.375
C = 59.875

MATLAB:

>> A = [57 59 60 100 59 58 57 58 300 61 62 60 62 58 57];

>> [TF, L, U, C] = isoutlier(A, 'quartiles')

TF =

  1×15 logical array

   0   0   0   1   0   0   0   0   1   0   0   0   0   0   0


L =

   52.3750


U =

   67.3750


C =

   59.8750

[pr0m1th3as]

You are correct about dim in quantiles method. I also get your approach in calling the function twice, since it eases the burden of handling dim > 1 later on, when data is a matrix or an ND-array. However, the same shape-related bug is still present in the percentiles_method. It reads L = Q(1)andU = Q(2)out ofquantile (x, P, dim)`, which is linear indexing into a result that only has two elements when the input is a vector, so every matrix silently gets the wrong thresholds instead of an error.

On another item: your changes purge a @end itemize texinfo command which results in help rendering plain text instead of texinfo. Make sure that you always run help <name> after applying any changes to make sure that help docstrings are rendered correctly.

Also it helps if you stamp the MATLAB run with version (this is a more general guide rather than specific to this PR).

Another thing I noticed in the function (unrelated to your PR, but you are welcome to fix it along) is that all methods that require a second input argument for window do not check if varargin{2} exists before trying to parse it into window = varargin{2};, which will error out without a proper error message. It would be nice to do the check and of course add the BISTs that verify that input parsing works properly.