support parsing dictionary formats

[rhdunn]

The document reader should have the ability to parse dictionary document formats and generate reader events for them. This should support:

• dictionary entries -- mapping (grapheme, language) to (pronunciation, phoneme-format, context/part-of-speech) or (say-as, context/part-of-speech);
• rule entries -- mapping (left, match, right, language) to (pronunciation, phoneme-format).

Where phoneme-format can be "ipa", "kirshenbaum", "ascii-ipa", "sampa", "x-sampa", "espeak", etc.

This should support the following (and other) dictionary formats:

• PLS (Pronunciation Lexicon Specification);
• eSpeak text;
• eSpeak binary;
• Cainteoir Dictionary Format;
• CMU (Carnegie Mellon University) Pronouncing Dictionary (low priority);
• OALED electronic dictionary (low priority).

This allows tools to be built on top of this support. For example:

• pronunciation verification -- simplifying tests/dictionary.py;
• loading user dictionaries into the engine when running;
• loading document dictionaries into the engine (e,g, ePub 3.0 with an embedded PLS document) before reading.

--- Want to back this issue? **[Post a bounty on it!](https://www.bountysource.com/issues/1026797-support-parsing-dictionary-formats?utm_campaign=plugin&utm_content=tracker%2F254961&utm_medium=issues&utm_source=github)** We accept bounties via [Bountysource](https://www.bountysource.com/?utm_campaign=plugin&utm_content=tracker%2F254961&utm_medium=issues&utm_source=github).