Schemas cleanup: Handle IntNodes properly

IntNodes were getting caught by the StandardAnalyzer for small
values. This could probably be avoided by using whoosh.fields.INT
in the schema declaration as well (and that should get done
eventually). However, a SimpleAnalyzer as the default is better for
l2cs' case anyway, because the CloudSearch server will handle
stop words and query optimizations.

Additionally, IntNodes are now properly generating unicode
elements.

Fix issue #18: IntNodes need to use unicode

Schema usage requires unicode. This changeset brings
IntNodes in-line with that necessity.

Issue #1: Add support for whoosh Schemas

Note that when using a Schema, the parser
will require Unicode objects to function

Revert "Fix issue #16: Incorrect quoting of phrases"

This reverts commit 76eff93.

The "fix" creates two meanings for double-quotes:
* encapsulating a query against a field, e.g.: foo:"bar baz"
* grouping exact phrases, e.g.: "jon smith"

The two must be disambiguated to avoid confusion or loss
of features

Fix issue #16: Incorrect quoting of phrases

Fix incompatible FnTagger call

A call to FnTagger used syntax that was
only available in whoosh version 2.4+. This
change reverts to the 2.3 compatible syntax.

Fix Issue #11: KeyError on NullQuery

NullQuery clauses were being generated upon
encountering representations of empty strings, e.g.

    foo:""

This commit adds a handler for NullQuery clauses

Further import issues with setup.py

Fix import issue in setup.py

Bumps version to 1.0.3

Set version to 1.0.2