Luqum is using the global lexer by default that may conflict with other libraries using ply

[Morikko]

Problem

import luqum.parser

lucene_query = "field: value"

luqum.parser.parser.parse(lucene_query)
# No error

import plyara

luqum.parser.parser.parse(lucene_query)
# No error

plyara.Plyara()

luqum.parser.parser.parse(lucene_query)
# Error: luqum.exceptions.ParseSyntaxError: Syntax error in input : unexpected  'field' at position 0!

As opposed to #72, the example is failing in a single thread mode.

Also using luqum.thread.parse(lucene_query) may still fail depending of the ordering:

1. Before plyara.Plyara(): No error as the new global lexer was not changed yet
2. After plyara.Plyara(): Same error as lex.lexer.clone() is used instead of luqum.parser.lexer

Cause

luqum is defining a global lexer in https://github.com/jurismarches/luqum/blob/master/luqum/parser.py#L248 that is used "magically" by PLY.

The PLY documentation has a section Multiple Parsers and Lexers for multiple "lexer" usage.

Workaround

luqum.parser.parser.parse(raw_lucene_query, lexer=luqum.parser.lexer)`

Idea

1. luqum.thread.parse should use luqum.parser.lexer instead of lex.lexer.clone(). Also the lexer parameter in luqum.thread.parse is not even usable as a workaround.

2. Provide a parser interface that automatically use the luqum.parser.lexer. Although, there is a workaround, I still believe the user should not have to know PLY internal in order to use luqum in a global environment where other libraries may use PLY. It is the solution implemented by plyara.Plyara, so plyara is still parsing correctly even if luqum.parser would be imported first.

[alexgarel]

@Morikko it's a good idea ! Do you want to champion to propose a PR ?