Trigger CI on PR, as well as push

[Xophmeister]

Also updated benchmark to use new tree-sitter-ocaml API.

[Xophmeister]

There are still a few broken things... I will fix and resubmit for review.

[Xophmeister]

The problems appear to be with the WASM target, from the changes introduced in #794, if I'm following the log output correctly. It complains about a bunch of stuff in topiary-core/src/tree-sitter.rs:

• tree_sitter_facade::QueryMatch is a trait for native code, but it's a struct for WASM. It needs to be imported for native, but in WASM we get a dead code warning.
• tree_sitter_facade::Query::matches returns a streaming_iterator::StreamingIterator implementation in native code, hence needing the trait imported, but an ExactSizeIterator implementation in WASM. Another dead code warning.
• tree_sitter_facade::QueryMatch::pattern_index returns a usize in native code and a u32 in WASM, so we get downstream type mismatch errors.
• From<tree_sitter::Point> is not implemented for topiary_tree_sitter_facade::Point for WASM. (From<topiary_web_tree_sitter_sys::Point> is implemented instead, in WASM.)

[Xophmeister]

• From<tree_sitter::Point> is not implemented for topiary_tree_sitter_facade::Point for WASM. (From<topiary_web_tree_sitter_sys::Point> is implemented instead, in WASM.)

I'm not sure how to solve this one. If I add this to the WASM module in topiary-tree-sitter-facade/src/point.rs:

    impl From<tree_sitter::Point> for Point {
        fn from(point: tree_sitter::Point) -> Self {
            topiary_web_tree_sitter_sys::Point::new(point.row, point.column).into()
        }
    }

The error changes to:

error[E0433]: failed to resolve: use of undeclared crate or module `tree_sitter`
   --> topiary-tree-sitter-facade/src/point.rs:121:15
    |
121 |     impl From<tree_sitter::Point> for Point {
    |               ^^^^^^^^^^^ use of undeclared crate or module `tree_sitter`

This is true and by design...

[Xophmeister]

...fixed that by avoiding the conversion call when targetting WASM.

[aspiwack]

I don't think there's any reason not to run CI on pull requests which don't target master. Especially since there's this playground branch that I believe we still have, which is also in production.