Jason Hansen

When I run the benchmarks with the default options, it doesn't look like there's much of an improvement. ``` flamegraph/dtrace time: [991.36 us 992.91 us 994.62 us] thrpt: [76.382 MiB/s...

Here's the flame graph with the changes @bcmyers made.    [flamegraph.svg.zip](https://github.com/jonhoo/inferno/files/3006344/flamegraph.svg.zip)

I tried using `FnvHashMap` for frame merging, but it didn't seem to make a difference.

This is a differential flame graph comparing with and without `--reverse`.  [diff.svg.zip](https://github.com/jonhoo/inferno/files/3010358/diff.svg.zip)

Well both `flamegraph::filled_rectangle` and `svg::write_str` have around 20x the number of samples when using `--reverse`. | Function | Samples | Samples `--reverse` | | :--- | :--- | :--- |...

I benchmarked with `hashbrown` + `fnv`, but no change in performance was detected.

@jonhoo I wouldn't be opposed to that idea.

FWIW the current implementation matches the output of [FlameGraph](https://github.com/brendangregg/FlameGraph), and from what I can tell it's the intended behavior. See the part boxed in red below from [differential-flame-graphs](https://www.brendangregg.com/blog/2014-11-09/differential-flame-graphs.html). 

Here's what I did to disable in telescope: ```lua require('pears').setup(function(conf) conf.disabled_filetypes { '' } end) ```

I don't know why disabling the 'TelescopePrompt` filetype doesn't work. That's actually what I tried at first, and then out of curiosity I tried it with an empty string and...