Very few of the compiler regression tests include a comment
saying waht they do. Many are obvious, some are anything but.
I've started with a-c in the top directory. More will follow once
we agree on the approach, correctness, and thoroughness here.
zerodivide.go sneaked in too.

R=rsc, r
CC=golang-dev
https://golang.org/cl/5656100

X ,s;^// \$G (\$D/)?\$F\.go *$;// compile;g
X ,s;^// \$G (\$D/)?\$F\.go && \$L \$F\.\$A *$;// build;g
X ,s;^// \$G (\$D/)?\$F\.go && \$L \$F\.\$A && \./\$A\.out *$;// run;g
X ,s;^// errchk \$G( -e)? (\$D/)?\$F\.go *$;// errorcheck;g

R=golang-dev, bradfitz
CC=golang-dev
https://golang.org/cl/5656082

Fixes #1316.

R=ken2
CC=golang-dev
https://golang.org/cl/3575042

This shortens, simplifies and regularizes the code significantly.
(Improvements to reflect could make another step.)
Passes all.bash.

One semantic change occurs: The String() method changes
behavior. It used to run only for string formats such as %s and %q.
Instead, it now runs whenever the item has the method and the
result is then processed by the format as a string. Besides the
regularization, this has three effects:

	1) width is honored for String() items
	2) %x works for String() items
	3) implementations of String that merely recur will recur forever

Regarding point 3, example from the updated documentation:
	type X int
	func (x X) String() string { return Sprintf("%d", x) }
should cast the value before recurring:
	func (x X) String() string { return Sprintf("%d", int(x)) }

R=rsc
CC=golang-dev
https://golang.org/cl/1613045

in the tests, println+panic.
gofmt some tests too.

R=rsc
CC=golang-dev
https://golang.org/cl/741041

various bug fixes and tests involving constants.

test/const1.go is the major new test case.

R=ken
OCL=26216
CL=26224