Add phan

Bug: T303786
Change-Id: I8a8bfdf5bb0c116db3817bdd1aea442f0080ff93

Drop PHP support pre 7.0

Pin testing to hhvm-3.18

Change-Id: I097b9391620718bc19963411734075b6968b5c80

Update PHP TextCat Models to 10K n-grams

Update all LM/ and LM-query/ models to 10K n-grams. The number of spaces
('_') counted in the LM models has gone down by 2 for every model, but
doesn't change the rank statistics for any model.

Update lm2php.php to handle slightly changed Perl model format (a stray
space was removed).

Add a couple of test cases that differ by model size above 5K (previous
max).

Bug: T155672
Change-Id: If35912574e833a677459531f994ae95f314b042d

Add newly validated query-based language models to TextCat

Add validated models for language ID of queries that are not
already present. Includes Czech, Indonesian, Italian, Japanese,
Dutch, Polish, Portuguese, Swedish, Turkish, Ukrainian, and
Vietnamese

Bug: T121539
Change-Id: I44cb67fe411de32c9b0848058ef18cc95e83231f

Add autoload definition

Change-Id: I6c3108bcf4a1db05e3d8080bdeab016e6a24b238

Add .gitattributes file to remove unnecssary files from package

Change-Id: I854887acbb6a64bec51753a63fd2b63d46613366

Create Wiki-Text-based language models for TextCat

Moved existing query-based models to LM-query/. Created 70 new models
based on random articles from the relevant Wikipedia. Minor updates
to PHP code, including change output join text to "OR" so as not to
conflict with language model "or.lm". Major updates to README.md.

These models have not been evaluated (see T121539), but are made
available as is.

BUG: T121545

Change-Id: I772670f2fa97dfe3981fd139ea40c62f921ccda7

Code style fixes

Change-Id: I402c69f8d1a9f4d8bf3515d220c2ae612d9de404

Less randomness