remove inline and let the optimizer decide

Pretty much the same for WASM size (a few byte smaller without inlining
the wrappers at the call sites). Instead LLVM inlines the factored-out
function back into the wrappers so performance wise there's no
difference.

query-engine/query-structure/src/record.rs

@@ -182,7 +182,6 @@ impl Record {
     /// Extracts a `SelectionResult` from this `Record`.
     /// `field_names`: Database names of the fields contained in this `Record`.
     /// `selected_fields`: The selection to extract.
-    // #[inline]
     pub fn extract_selection_result_from_db_name(
         &self,
         field_names: &[String],
@@ -193,7 +192,6 @@ impl Record {
 
     /// Extracts a `SelectionResult` from this `Record` using Prisma field names rather than
     /// database names.
-    // #[inline]
     pub fn extract_selection_result_from_prisma_name(
         &self,
         field_names: &[String],