[ML] FileViz - allow customisation of the number of lines to analyze

[droberts195]

At present the file data visualizer always tells the backend endpoint to analyze the first 1000 lines of the uploaded file.

This is a reasonable default for the first analysis. However, the user might want to see field statistics for the whole file, so on the second and subsequent calls to the backend endpoint the UI should allow the number of lines to sample to be set to a different number. (The absolute minimum should be 3. The maximum could be 1000000 or 999999, because any number greater than the number of lines in the uploaded file means "use the whole file".)

[elasticmachine]

Pinging @elastic/ml-ui