feat(ocamllsp): support Pexp_ifthenelse folding ranges

[tatchi]

Support folding ranges for ifthenelse expressions.

Before:

Screen.Recording.2023-02-15.at.06.50.12.mov

After:

Screen.Recording.2023-02-15.at.06.50.32.mov

As you can see in the video above, we can collapse the entire if expression but not the else. This is because we don't have the start location of the else keyword. We only have the location of the whole expression inside the else. We can add this location to the ranges, but it doesn't appear next to the else.

Screen.Recording.2023-02-15.at.06.52.38.mov

Also, it makes it impossible to collapse only the first expression inside the else.

[rgrinberg]

Update CHANGES please

[rgrinberg]

@voodoos do you want to rebase and merge this?

[coveralls]

Pull Request Test Coverage Report for Build 4088

• -4 of 4 (0.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.003%) to 20.381%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
ocaml-lsp-server/src/folding_range.ml	0	4	0.0%

Totals
Change from base Build 4082:	-0.003%
Covered Lines:	4851
Relevant Lines:	23801

---

💛 - Coveralls

[voodoos]

@rgrinberg are you familiar with this CI failure ?

--- a/_build/default/ocaml-lsp-server/test/e2e-new/workspace_change_config.ml
+++ b/_build/default/ocaml-lsp-server/test/e2e-new/workspace_change_config.ml.corrected
@@ -68,7 +68,35 @@ let string = "Hello"
   in
 
   test source req;
-  [%expect {| CodeLens found: 0 |}]
+  [%expect {|
+    (* CR expect_test_collector: This test expectation appears to contain a backtrace.
+       This is strongly discouraged as backtraces are fragile.
+       Please change this test to not include a backtrace. *)
+
+    CodeLens found: 0
+    detached: /-----------------------------------------------------------------------
+    | Internal error: Uncaught exception.
+    | ("Lev_fiber.Io: already closed",
+    | { source =
+    |     "Raised by primitive operation at Lev_fiber.Io.callstack.(fun) in file \"submodules/lev/lev-fiber/src/lev_fiber.ml\", line 590, characters 31-58\n\
+    |      Called from Lev_fiber.Io.create in file \"submodules/lev/lev-fiber/src/lev_fiber.ml\", line 743, characters 17-29\n\
+    |      Called from Lev_fiber.Io.Lazy_fiber.force.(fun) in file \"submodules/lev/lev-fiber/src/lev_fiber.ml\", line 1047, characters 23-27\n\
+    |      Called from Fiber__Scheduler.exec in file \"fiber/src/scheduler.ml\", line 73, characters 8-11\n\
+    |      Called from Fiber__Scheduler.start in file \"fiber/src/scheduler.ml\", line 216, characters 2-42\n\
+    |      Called from Fiber.run.(fun) in file \"fiber/src/fiber.ml\" (inlined), line 17, characters 28-47\n\
+    |      Called from Lev_fiber.run in file \"submodules/lev/lev-fiber/src/lev_fiber.ml\", line 1264, characters 10-59\n\
+    |      Called from Ocaml_lsp_server.run in file \"ocaml-lsp-server/src/ocaml_lsp_server.ml\", line 936, characters 2-151\n\
+    |      Called from Stdune__Exn_with_backtrace.try_with in file \"otherlibs/stdune/src/exn_with_backtrace.ml\", line 9, characters 8-12\n\
+    |      Called from Dune__exe__Main in file \"ocaml-lsp-server/bin/main.ml\", line 41, characters 6-106\n\
+    |      "
+    | })
+    | Raised at Stdune__Code_error.raise in file "otherlibs/stdune/src/code_error.ml", line 10, characters 30-62
+    | Called from Lev_fiber.Io.with_ in file "submodules/lev/lev-fiber/src/lev_fiber.ml", line 983, characters 10-62
+    | Called from Fiber__Scheduler.exec in file "fiber/src/scheduler.ml", line 73, characters 8-11
+    | Re-raised at Stdune__Exn.raise_with_backtrace in file "otherlibs/stdune/src/exn.ml" (inlined), line 38, characters 27-56
+    | Called from Stdune__Exn_with_backtrace.reraise in file "otherlibs/stdune/src/exn_with_backtrace.ml", line 20, characters 33-71
+    | Called from Fiber__Scheduler.exec in file "fiber/src/scheduler.ml", line 73, characters 8-11
+    \----------------------------------------------------------------------- |}]

[rgrinberg]

Yes, I've seen them before. Our tests aren't very careful about making sure that we close the client only after we receive or reject incoming packets from the server. I intend to fix this eventually, but it's only an issue that's observable in the tests (nobody spins up lsp servers to serve a single request) so priority is low