There are two points that affect performance in the current json load:
1 json is actually a traversal operation for searching by key. We call HasMember and operator[] twice, and need to traverse twice
2 If you use a string as the key to find the json object, strlen will be called every time, so the number of calls to strlen is the row number * column number

we have two way to improve this:
1 use FindMember instead of HasMember + []
2 add a flag of fuzzy_parse, if the json file all object keys are the same and has same order, we only need to parse the first row, and then use index instead key to parse value