Typically for a cnn architecture, in a single filter as described by your number_of_filters parameter, there is one 2d kernel per input channel.
There are input_channels *.
Why would cnn-lstm be another name for rnn, when it doesn't even have rnn in it?
Can you clarify this?
What is your knowledge of rnns and cnns?
Do you know what an lstm is?
You can use cnn on any data, but it's recommended to use cnn only on data that have spatial features (it might still work on data that doesn't have spatial features, see duttaa's comment.