Mark Differentiation related FloatingPoint methods with @inlinable for more optimization opportunities (#78806)

Co-authored-by: Jaap Wijnen <jaap@passivelogic.com>

Author: JaapWijnen

stdlib/public/Differentiation/FloatingPointDifferentiation.swift.gyb

@@ -42,6 +42,7 @@ extension ${Self}: Differentiable {
   public typealias TangentVector = ${Self}
 
   ${Availability(bits)}
+  @inlinable
   public mutating func move(by offset: TangentVector) {
     self += offset
   }
@@ -95,15 +96,15 @@ extension ${Self} {
 /// Derivatives of standard unary operators.
 ${Availability(bits)}
 extension ${Self} {
-  @usableFromInline
+  @inlinable
   @_transparent
   @derivative(of: -)
   static func _vjpNegate(x: ${Self})
     -> (value: ${Self}, pullback: (${Self}) -> ${Self}) {
     return (-x, { v in -v })
   }
 
-  @usableFromInline
+  @inlinable
   @_transparent
   @derivative(of: -)
   static func _jvpNegate(x: ${Self})