cell2table behaviour on numerical arrays of different sizes

This isn't actually a question. Rather, this is a feature request.

In the documentation of cell2table , it reads "If the cell contents have compatible sizes and types, then cell2table vertically concatenates the contents of the cells in each column of C to create each variable in T. If the cell contents have different sizes or incompatible types, then the corresponding variable in the table, T, is a column of cells."

K>> A = {[1 2 3 4 5],2,[1 2 3];...
    [6 7 8 9],5,4}

A = 

    [1x5 double]    [2]    [1x3 double]
    [1x4 double]    [5]    [         4]

K>> B = cell2table(A)

B = 

         A1         A2         A3     
    ____________    __    ____________

    [1x5 double]    2     [1x3 double]
    [1x4 double]    5     [         4]

As you can see, cell2table does what it promised to do, but it's not quite good for me. Say, the cell array A holds the results of computations and sample number for each cell of A is different. I believe this is not rare practice. So, I would like to see something like this:

B2 = 

         A1         A2         A3     
    ____________    __    ____________

    [1x5 double]    [2]     [1x3 double]
    [1x4 double]    [5]     [         4]

Otherwise, depending on the sample sizes, the data type of variables can be numeric or cell, making code highly error-prone. In this example, variables in A2 are just happened to be a scalar, and thus it is stored as double in B. It's kind of unpredictable. This turned out to be a major problem, when I replaced a cell array in my code with a table by using cell2table for better readability.

Of course I can "manually" amend B after cell2table (in fact I had to), by checking the class of resultant variables (columns). But wouldn't it be useful if there were an optional parameter/value pair that defines the behaviour of cell2table for all the columns or for each column?

I can think of a syntax like this:

K>> B3 = cell2table(A,'NonuniformOutput',true)

which returns all the columns as cell arrays

or

K>> B4 = cell2table(A,'NonuniformOutput',[false,true,true])

where you can specify the action column by column manner. I think I can write a wrapper function of cell2table to achieve this, but just thought, if it is officially implemented, it's better for many.