Clarify FRONT_BUFFER requirements

    /* Bits 28-31 are reserved for vendor usage */

    /**

    * Buffer is used for front-buffer rendering

    * Buffer is used for front-buffer rendering.

    *

    * To satisfy an allocation with this usage, the resulting buffer

    * must operate as equivalent to shared memory for all targets.

    *

    * For CPU_USAGE_* other than NEVER, this means the buffer must

    * "lock in place". The buffers must be directly accessible via mapping.

    *

    * For GPU_RENDER_TARGET the buffer must behave equivalent to a

    * single-buffered EGL surface. For example glFlush must perform

    * a flush, same as if the default framebuffer was single-buffered.

    *

    * For COMPOSER_* the HWC must not perform any caching for this buffer

    * when submitted for composition. HWCs do not need to do any form

    * of auto-refresh, and they are allowed to cache composition results between

    * presents from SF (such as for panel self-refresh), but for any given

    * present the buffer must be composited from even if it otherwise appears

    * to be the same as a previous composition.

    */

    FRONT_BUFFER                       = 1L << 32,