Improve transaction handling in Base1PluginExecuter

[Nicklas Nordborg]

The Base1PluginExecuter currently opens and closes transactions in some strange places. The first transactions span the export, execution and data import. It uses a second parallel transaction for attaching files during the data import. A third transaction is then used for some clean-up tasks like removing stdin and stdout.

It is definitely not needed to keep a transaction open during the execution phase. The first transaction should be closed after the data export. A second transaction should be used for data import. If possible things should be rearranged so there is no need for a parallel transaction for the file import or for the third transaction used for clean-up.

[Martin Svensson]

(In [4606]) Fixes #1152 Improve transaction handling in Base1PluginExecuter. There is only one transaction open each time and there are three separate transaction now: one for exporting data, one to import files and the last to import data and do some clean-up.

[Nicklas Nordborg]

(In [5773]) References #1152: Handling short read transcript sequence data

• Adds a 'Cufflinks' raw data type with five columns: coverage, fpkm, fpkm_lo, fpkm_hi and status
• Define FPKM_TRACKING file type and MIME type.
• Adds left(string, index) as a JEP formula so that we can parse out the chromosome from the 'locus' column in the tracking files.
• Define two new configurations for the raw data importer that parses cufflinks isoform files.