Add fast-path for simple filters

I have a ~600MB NDJSON file with ~1M lines (happy to share if it's helpful).

Using `jq` to search for an individual entry:

```
$ time jq 'select(.id==3260839)' file.ndjson
# ~9 seconds
```

Using `fx -s` to do the same is ~4x slower:

```
$ time fx -s file.ndjson '?.id == 3260839'
# ~33s
```

Using `skip` (ala #292) without `-s` is considerably faster but still slower than `jq`:

```
$ time fx file.ndjson 'x.id == 3260839 ? x : skip'
# ~18s
```

Of course, I can `grep` first to speed things up, but it would be nice if this weren't necessary:

```
$ time grep 3260839 file.ndjson | fx -s '?.id == 3260839'
# ~7s
```

Since there's only ~1M rows in this ndjson file, I doubt that running `x => x.id == 3260839` is the bottleneck. Maybe it's just the JSON parsing?