Should we make querystrings reserved words?

I think maybe we shouldn't to keep a simple schema, but we should do a big warning about this in the docs. We can do some crazy things with them:

```json
$ echo '{"data": {"_limit": 5}}' | http put localhost:8888/v1
{
"data": {
"_limit": 5,
"id": "b1",
"last_modified": 1482506056818
},
"permissions": {
"write": [
"basicauth:902db65276f24f6f084cd57fddb1bb3835054d27858a606f59410e8c227465e2"
]
}
}

$ http get localhost:8888/v1/buckets?_limit=1 -a a:a
{
"data": [
{
"_limit": 5,
"id": "b1",
"last_modified": 1482506056818
}
]
}
```

```json
$ echo '{"data": {"min_field": "10", "field": 5}}' | http put localhost:8888/v1/buckets/b1 -a a:a
{
"data": {
"field": 5,
"id": "b1",
"last_modified": 1482505784675,
"min_field": 10
},
"permissions": {
"write": [
"basicauth:902db65276f24f6f084cd57fddb1bb3835054d27858a606f59410e8c227465e2"
]
}
}

$ http get localhost:8888/v1/buckets?min_field=10 -a a:a
{
"data": []
}
```

No files, tests, or entry points are named. Start by tracing how query-string parameters are handled in the API and reviewing the documentation around field names; done means reaching an agreed policy on reserved parameters and documenting the warning or behavior clearly.

Written by the indexing model from the issue text.