Improve documentation

[pynchia]

Stellar works by storing copies of the database in the RDBMS (named as 
stellar_xxx_master and stellar_xxxx_slave). When restoring the database,
Stellar simply renames the database making it lot faster than the usual
SQL dump

Please explain better how it works, i.e. how the master and the slave are used.
Most of all, please explain if the copy (snapshot) invalidates when renamed, i.e. if you can take a snapshot and use it multiple times (useful in unit tests).
From the description one would think the copy is renamed, i.e. it becomes the desired DB, i.e. it will be overwritten when used, preventing it from being used again.
Poor documentation hampers adoption.