Implement a basic convolution operator

The goal of this issue is to add an initial implementation of a convolution operation that will be used to express most of the existing operators using a common computational kernel.

The convolution operation will be used for finite difference and needs to work in 1D only

The requirements are as follows:
- It will take a `BandMatrixRow` that will describe a convolution kernel
- It will take some kind of abstract array of numbers as the operands
- The function applies the standard discrete convolution
- It needs to support 'padding' the values to account for boundary conditions

It should be implemented as in a seperate file in `src/Operators` folder.

This is kind of a smaller subtask me and @sjavis though will be necessary to have after the discussion yesterday.

@imreddyTeja @dennisYatunin @petebachant If you have a minute to have a look in case we are going a bit off track.