`org-float' is now obsolete

* org-agenda.el (org-float): Don't alias `org-float'.

* org.texi (Timestamps, Deadlines and scheduling): Use
`diary-float' instead of the now obsolete alias `org-float'.

* test-org-element.el
(test-org-element/timestamp-interpreter): Ditto.

doc/org.texi

@@ -5520,7 +5520,7 @@ example with optional time
 
 @example
 * 22:00-23:00 The nerd meeting on every 2nd Thursday of the month
-  <%%(org-float t 4 2)>
+  <%%(diary-float t 4 2)>
 @end example
 
 @item Time/Date range
@@ -5858,7 +5858,7 @@ entries.  Org mode will issue early and late warnings based on the
 assumption that the timestamp represents the @i{nearest instance} of
 the repeater.  However, the use of diary sexp entries like
 @c
-@code{<%%(org-float t 42)>}
+@code{<%%(diary-float t 42)>}
 @c
 in scheduling and deadline timestamps is limited.  Org mode does not
 know enough about the internals of each sexp function to issue early and

lisp/org-agenda.el

@@ -5521,7 +5521,6 @@ Do we have a reason to ignore this TODO entry because it has a time stamp?
   (org-no-warnings
    (let ((calendar-date-style 'european)	(european-calendar-style t))
      (diary-date day month year mark))))
-(defalias 'org-float 'diary-float)
 
 ;; Define the` org-class' function
 (defun org-class (y1 m1 d1 y2 m2 d2 dayname &rest skip-weeks)

testing/lisp/test-org-element.el

@@ -2074,8 +2074,8 @@ CLOSED: [2012-01-01] DEADLINE: <2012-01-01> SCHEDULED: <2012-01-01>\n"))))
 		  "[2012-03-29 16:40]--[2012-03-29 16:41]")
 		 "[2012-03-29 16:40]--[2012-03-29 16:41]\n"))
   ;; Diary.
-  (should (equal (org-test-parse-and-interpret "<%%org-float t 4 2>")
-		 "<%%org-float t 4 2>\n"))
+  (should (equal (org-test-parse-and-interpret "<%%diary-float t 4 2>")
+		 "<%%diary-float t 4 2>\n"))
   ;; Timestamp with repeater interval.
   (should (equal (org-test-parse-and-interpret "<2012-03-29 +1y>")
 		 "<2012-03-29 +1y>\n")))