[meta] IE9 test fixes

[bill]

Meta ticket to cover test fixes for IE9, or workarounds that we add to test files. Actual IE9 fixes to the code should be in separate tickets.

[bill]

In [27120]:

IE9 chokes on new Function("return"), stemming from onClick="return". Since this is arguable a syntax error anyway, updating the test file to have a semicolon after return, and after other statements too. Refs #12430, #14322.

[bill]

In [27129]:

print out text of Error (non doh._AssertFailure Error) rather than [Object Error], refs #14322 !strict

[bill]

In [27130]:

[24328] erased changes from [24052], putting them back into trunk/, refs #12463, #7681, #14322 !strict.

[bill]

In [27131]:

[24328] erased changes from [24052], putting them back into 1.7/ branch, refs #12463, #7681, #14322 !strict.

[bill]

In [27134]:

eliminate race conditions waiting for Trees to initialize, refs #14322

[bill]

In [27135]:

More Tree test fixes, refs #14322 !strict.

[bill]

In [27136]:

On IE9 Error.toString() return "[Object Error]", not the text of the error, refs #14322

[bill]

In [27137]:

workaround problem during robot tests where addOnWindowUnload() called twice, refs #14322 !strict

[bill]

In [27140]:

Tree_a11y.html needs a longer timeout waiting for trees to load (on my IE8 machine), refs #14322 !strict

[bill]

In [27147]:

increase timeout, refs #14322

[bill]

In [27148]:

increase timeout, refs #14322

[bill]

In [27150]:

more test fixes, refs #14322

[bill]

In [27152]:

increase timeout, refs #14322

[bill]

In [27189]:

leverage waitForLoad, refs #14322

[bill]

In [27286]:

Merge [27129] to 1.7/ branch: print out text of Error (non doh._AssertFailure Error) rather than [Object Error], refs #14322 !strict

[bill]

In [29421]:

IE9 gets confused by an empty return value, so return some text, refs #14322

[Colin Snover]

1.8 has been tagged; moving all outstanding tickets to next major release milestone.

[bill]

IE9 tests are working fairly well now, and the remaining issues (quirksmode tests are running in standards mode) are filed as separate tickets.

[haysmark]

I don't think it's essential to backport these test fixes.