Split files when planning scan tasks

[rdblue]

When building a scan, the TableScan API can plan the files to read (planFiles) or group the files into combined splits (planTasks). Split planning should also split files at the target split size before bin packing to create the final splits.

This relates to adding split locations to the manifest file (row group or stripe offsets). The simple version of this issue is to split at the target split size and then combine, but eventually we want to take the split offsets into account if it does make sense to store them in the manifest file.

[rdblue]

Fixed in #111.