fix issue when response contains ANSI escape sequences

[dmitryk-dk]

There is a parsing issue with ANSI escape sequences, which a standard JSON decoder can't handle.
I used valyala fastjson library to solve this issue. Victoria logs can handle logs like
{"_time":"2024-06-26T13:15:15.000Z","_stream_id":"00000000000000009eaf29866f70976a098adc735393deb1","_stream":"{compose_project=\"app\",compose_service=\"gateway\"}","_msg":"\x1b[2m2024-06-26T13:15:15.004Z\x1b[0;39m \x1b[32mTRACE\x1b[0;39m \x1b[35m1\x1b[0;39m \x1b[2m---\x1b[0;39m \x1b[2m[ parallel-19]\x1b[0;39m \x1b[36mo.s.c.g.f.WeightCalculatorWebFilter \x1b[0;39m \x1b[2m:\x1b[0;39m Weights attr: {} ","compose_project":"app","compose_service":"gateway"}

where \x1b[0;39m can be present.

Related issue: #24

[hagen1778]

@dmitryk-dk do I understand it right that #24 is fixed on the vmlogs backend side and no changes in the plugin are required?

[dmitryk-dk]

@dmitryk-dk do I understand it right that #24 is fixed on the vmlogs backend side and no changes in the plugin are required?

Yeah, I discussed the problem with @valyala, and he said that the main problem is on the Victorialogs side. But this PR helps to handle both situations with ANSI escape sequences and Unicode chars in the strings.
I have added tests for both situations, and they work well.
So, i think it would be helpful to have changes in our datasource