Namespace Pillar Package Table

[raizyr]

It seems wrong to me that a formula specific package_table should be in the pillar root level. This pull request moves it to live with the rest of the pillar data expected by the formula in pillar['salt']. Additionally, I removed the if/elif statement in favor of pillar.get, this will provide simple fallback to the hardcoded package_table if the os doesn't exist in the pillar package_table.

[wwentland]

To be honest I find the package_table in pillar to be quite unconventional too and would have rather expected this to be handled like in other formulas by defining various dictionaries/maps in map.jinja that hold the package data and that will use merge=salt['pillar.get']('salt:lookup') or variations thereof.

Would you, @raizyr, be interested in changing the code in this PR to adhere more closely to the conventions defined in http://docs.saltstack.com/en/latest/topics/development/conventions/formulas.html ? It would also require rebasing against master.

There might, however, be a specific reason for this design that escapes me right now.

[whiteinge]

I second @BABILEN 's suggestion to follow other formula conventions for this with salt:lookup.

[gravyboat]

@raizyr Are you planning on making the changes with salt:lookup? Otherwise I'd like to close this PR.

[aboe76]

@raizyr please look at #81 if this better?

[aboe76]

@raizyr and @nmadhok can this pull request in because #81 got merged.