Optimize parser by using nested parser state

[zuiderkwast]

When parsing a nested array in chunks efficiently, the parser state
(continuation data) needs to be a nested structure. The parser is
restructured.

Fixes wooga#127

[bjosv]

LGTM, how is the cprof runs as in wooga#127 (comment) turning out with this change?

[zuiderkwast]

LGTM, how is the cprof runs as in wooga#127 (comment) turning out with this change?

Erlang/OTP 23 [erts-11.1.8] [source] [64-bit] [smp:12:12] [ds:12:12:10] [async-threads:1]

Eshell V11.1.8  (abort with ^G)
1> {ok, E} = eredis:start_link().     
{ok,<0.233.0>}
2> eredis:q(E,["FLUSHDB"]).
{ok,<<"OK">>}
3> A = binary:copy(<<"a">>, 10000000).
<<"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"...>>
4> eredis:q(E,["HSET", "h", "1", A]).
{ok,<<"1">>}
5> cprof:start(), R=eredis:q(E, ["HGETALL", "h"], 100000), cprof:pause(),R.
{ok,[<<"1">>,
     <<"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"...>>]}
6> cprof:analyse(eredis_parser).
{eredis_parser,20571,
               [{{eredis_parser,do_parse,2},6860},
                {{eredis_parser,return,2},6854},
                {{eredis_parser,parse,2},6854},
                {{eredis_parser,split_by_newline,2},3}]}
7> cprof:stop().
15886
8> timer:tc(fun() -> eredis:q(E, ["HGETALL", "h"], 100000) end).
{21970,
 {ok,[<<"1">>,
      <<"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"...>>]}}
9> timer:tc(fun() -> eredis:q(E, ["HGET", "h", "1"]) end).      
{20421,
 {ok,<<"aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"...>>}}

Compared to wooga/eredis, there are fewer function calls and is faster also for the fast example. Tests done on my laptop. Times in microseconds.

Command	wooga/eredis	This PR
HGET h 1	25117	20421
HGETALL h	9429598	21970

[bjosv]

Nice