Improve type inference for @something

`@something` eagerly unwraps any `Some` given to it,
while keeping the variable between its arguments the same.
This can be an issue if a previously unpacked value is used as input
to `@something`, leading to a type instability on more than two arguments
(e.g. because of a fallback to `Some(nothing)`). By using different
variables for each argument, type inference has an easier time handling
these cases that are isolated to single branches anyway.

base/some.jl

@@ -138,10 +138,31 @@ true
     This macro is available as of Julia 1.7.
 """
 macro something(args...)
-    expr = :(nothing)
+    noth = GlobalRef(Base, :nothing)
+    something = GlobalRef(Base, :something)
+
+    # This preserves existing semantics of throwing on `nothing`
+    expr = :($something($noth))
+
+    #=
+    We go through the arguments in reverse
+    because we're building a nested if/else
+    expression from the inside out.
+    The innermost thing to check is the last argument,
+    which is why we need the last argument first
+    when building the final expression.
+    =#
     for arg in reverse(args)
-        expr = :(val = $(esc(arg)); val !== nothing ? val : ($expr))
+        val = gensym()
+        expr = quote
+            $val = $(esc(arg))
+            if !isnothing($val)
+                # unwrap eagerly to help type inference
+                $something($val)
+            else
+                $expr
+            end
+        end
     end
-    something = GlobalRef(Base, :something)
-    return :($something($expr))
+    return expr
 end