Fix jackknife MATLAB compatibility

[sahilphad07-sudo]

jackknife did not correctly handle multiple valid estimator outputs and edge cases. Estimators returning multiple values caused a dimension mismatch, while empty and single-element inputs produced results that differed from MATLAB.

This change updates jackknife to correctly handle multi-output estimators and the empty and single-element input cases.

Before

octave:1> jackknife (@(x) [mean(x); std(x)], [1 2 3])
error: =: nonconformant arguments (op1 is 1x1, op2 is 2x1)

octave:2> jackknife (@mean, [])
ans = [](0x1)

octave:3> jackknife (@mean, 5)
ans = NaN

After

octave:1> jackknife (@(x) [mean(x); std(x)], [1 2 3])

ans =

   2.5000   0.7071
   2.0000   1.4142
   1.5000   0.7071

octave:2> jackknife (@mean, [])
ans =

   NaN

octave:3> jackknife (@mean, 5)
ans =

   5

MATLAB Behavior

>> jackknife (@(x) [mean(x); std(x)], [1 2 3])

ans =

    2.5000    0.7071
    2.0000    1.4142
    1.5000    0.7071

>> jackknife (@mean, [])

ans =

   NaN

>> jackknife (@mean, 5)

ans =

     5

Regression Tests Added

• Empty input
• Single-element input
• Estimator returning multiple values

Test Results

All existing and new jackknife BISTs pass:

PASSES 4 out of 4 tests.