Add vec explicit conversion

This is change 2 of 9 that fix problems with the specification of the
`vec` class.  An implementation that follows the existing specification
would not accept common code patterns and would not pass the CTS.  None
of the existing implementations actually follow the existing
specification.

This change adds an explicit conversion when the `vec` has one element,
allowing a conversion to `T` whenever `DataT` can be explicitly
converted to `T`.

This enables code like the following, where the element type `DataT` is
itself a class:

```
sycl::vec<sycl::half, 1> h1;
int i = static_cast<int>(h1);
if (h1) {}
```

These changes correspond to slides 8 - 9 of the presentation that was
discussed in the WG meetings.

adoc/chapters/programming_interface.adoc

@@ -17869,6 +17869,21 @@ to which [code]#DataT# is implicitly convertible.
 Note that conversion operator shall not be templated
 to allow standard conversion sequence for implicit conversion.
 
+a@
+[source]
+----
+template<typename T>
+explicit operator T() const
+----
+a@ _Constraints:_
+
+* [code]#NumElements# is equal to 1, and
+* [code]#DataT# can be explicitly converted to [code]#T# via
+  [code]#static_cast<T>#, and
+* [code]#T# is not [code]#DataT#.
+
+Returns the value of the vector's element converted to [code]#T#.
+
 a@
 [source]
 ----
@@ -18661,6 +18676,10 @@ operator vector_t() const
 #endif
 
 operator DataT() const
+
+template<typename T>
+explicit operator T() const
+
 static constexpr size_t byte_size() noexcept
 static constexpr size_t size() noexcept
 

adoc/headers/vec.h

@@ -58,6 +58,10 @@ template <typename DataT, int NumElements> class vec {
   // Available only when: NumElements == 1
   operator DataT() const;
 
+  // Available only when: NumElements == 1
+  template<typename T>
+  explicit operator T() const;
+
   static constexpr size_t byte_size() noexcept;
 
   static constexpr size_t size() noexcept;