Pymamba

* Switching to PyPi
* You can now download this package with "pip", note however it will only work with Python3 for now.

* Fixed bug in empty to stop it deleting indecies
* Added 'save' method to table for updating pre-existing items (index aware)
* Added partial indexes, so only items where the key is not null are included in the index
* Code coverage hits 100%

* Added seek(index, record) - seek to the first matching record based on an index key

For example;

table.index('by_compound', '{cat}|{name}')
for doc in table.seek('by_compound', {'cat': 'A', 'name': 'Squizzey'}):
print(doc)


* Added range(index, lower, upper) - return all records with keys falling within the
limits set up (upper, lower)

For example;


table.index('by_compound', '{cat}|{name}')
for doc in table.range('by_compound', {'cat': 'A', 'name': 'Squizzey'}, {'cat': 'B', 'name': 'Gareth Bult1'):
print(doc)

* Simplified the way indexes work, maintaining a second 'easy' method is no
longer worthwhile with the new indexing interface. All indexes are now specified
using format string, so a simple index on a field called name is specified as;

- {name}

A compound index based on name and age would be;

- {name}{age}

Appling a little logic, if you want records in age order, and age is always in
the range 0-999 and is sorted numerically, we want the index specification to be;

- {age:03}{name}

i.e. a numberic, formatted as a string, length 3, zero padded

If we wanted sorting by name, we would probably want;

- {name}|{age:03}

Not the use of a sepatating character (|) to ensure separation between the variable
length alphanumeric key and the fixed length numeric key.

* Re-implemented 'find' method for Table
* Now returns a generator rather than a list
* Also accepts a filter, optionally applied after the index

* Implemented reindex method for Index class.
Indexing a table containing data will create an initial index from this data.