Reimplement JsonExpressionParser in terms of jsonparser

[MasslessParticle]

This PR reimplements the functionality of JSONExpressionParser using the jsonparser library rather than jsoniter

Benchmarks:

benchstat json_expression_old.txt json_expression_new.txt 
name                                    old time/op    new time/op    delta
JsonExpressionParser/json-expression-8    4.15µs ± 4%    0.82µs ± 3%  -80.17%  (p=0.000 n=10+10)

name                                    old alloc/op   new alloc/op   delta
JsonExpressionParser/json-expression-8    1.07kB ± 0%    0.02kB ± 0%  -97.76%  (p=0.000 n=10+10)

name                                    old allocs/op  new allocs/op  delta
JsonExpressionParser/json-expression-8      72.0 ± 0%       2.0 ± 0%  -97.22%  (p=0.000 n=10+10)

[MasslessParticle]

The index syntax for paths in jsonparser differs slightly from jsoniter. This function turns our representation into what is expected by the new implementation

[dannykopping]

Great work @MasslessParticle! Really excited for this. I'd be keen to see how this stacks up against the regular json parser; if it's now faster/more resource-efficient, we can start recommending its use when specific keys are needed.

Added a comment about error handling, but overall LGTM

[chaudum]

Are closing brackets really valid JSON start tokens?

[MasslessParticle]

Good catch, thanks! This is copy paste from the parser internals

[MasslessParticle]

@dannykopping Here's @cyriltovena's original PR for JSON Parser: #7723.

the jsonparser library helped somewhat. this PR ought to buy us quite a bit -- at lease for metrics queries.

[MasslessParticle]

@dannykopping It's interesting to note that when the JSON payload is small and doesn't contain nested keys, the JSON Parser performs about as well as logfmt. These benchmarks use a simpler JSON that @cyriltovena's