Fix derivatives in 2D-table for one-sided extrapolation by constant continuation

[beutlich]

Closes #3894.

@GarronFish You can test by cloning master branch of ModelicaTableAdditions where I already updated the binaries.

[GarronFish]

Closes #3894.

@GarronFish You can test by cloning master branch of ModelicaTableAdditions where I already updated the binaries.

I've tested the functionality using tabletest4.txt attached.
tabletest4.txt

The input signals go "around" the table 0,1,2
1,1,2
2.2,3
to test the derivative calculation for each case.
The test gave the expected results.

Functionally this appears to be working correctly now. Thanks!

[beutlich]

I've tested the functionality using tabletest4.txt attached.

Nice stimulation. I added it as new test model. Thanks.

Functionally this appears to be working correctly now. Thanks!

Thanks for testing and confirmation.

[GarronFish]

I've tested the functionality using tabletest4.txt attached.

Nice stimulation. I added it as new test model. Thanks.

Functionally this appears to be working correctly now. Thanks!

Thanks for testing and confirmation.

The test is better if the table is set so the derivative values are unique for example:
table=[0,1,2; 1,1,7; 2,2,4],

[HansOlsson]

I'm having a bit of trouble understanding how this is supposed to work. Basically I expected the getValue code to be matched by derivative code, but the codes are different and I only see how the new code matches linear interpolation in table and hold-extrapolation - but not for other interpolations for hold-interpolation. Am I missing something obvious?

[beutlich]

The change considers the cases to HOLD_LAST_POINT where there is one-sided extrapolation.

[HansOlsson]

Seems good now.