Optimize parser by using nested parser state

When parsing a nested array in chunks efficiently, the parser state
(continuation data) needs to be a nested structure. The parser is
restructured. It's also faster in other cases.

This fixes the issue in the original wooga/eredis:
wooga#127

include/eredis.hrl

@@ -25,8 +25,14 @@
 %% Continuation data is whatever data returned by any of the parse
 %% functions. This is used to continue where we left off the next time
 %% the user calls parse/2.
--type continuation_data() :: any().
--type parser_state() :: status_continue | bulk_continue | multibulk_continue | error_continue.
+-type continuation_data() ::
+        start |
+        {status_continue, Acc :: binary()} |
+        {error_continue, Acc :: binary()} |
+        {bulk_size, Acc :: binary()} |
+        {multibulk_size, Acc :: binary()} |
+        {bulk_continue, BytesLeft :: integer(), Acc :: binary()} |
+        {multibulk_continue, NumLeft :: integer(), Acc :: list()}.
 
 %% Internal types
 -ifdef(OTP_RELEASE). % OTP >= 21
@@ -42,8 +48,7 @@
 %% Internal parser state. Is returned from parse/2 and must be
 %% included on the next calls to parse/2.
 -record(pstate, {
-          state = undefined :: parser_state() | undefined,
-          continuation_data :: continuation_data() | undefined
+          states = [] :: [continuation_data()]
 }).
 
 -define(NL, "\r\n").