Fix spacing for JSCollection, JSSet, and JSMap.

Fix spacing for JSCollection, JSSet, and JSMap.

tools/gen-postmortem-metadata.py skips JSCollection, JSSet, and JSMap
due to the spacing in objects.h. This change allows them to be picked
up.

R=machenbach@chromium.org
BUG=

[Michael Achenbach, 2015-11-10 06:56:50]

machenbach@chromium.org changed reviewers:
+ jkummerow@chromium.org

[Michael Achenbach, 2015-11-10 06:56:51]

lgtm

[Jakob Kummerow, 2015-11-10 09:38:18]

ACK.

Semantic meaning of spaces is a horrible, horrible hack. I would appreciate it
if gen-postmortem-metadata.py could migrate to a saner approach, e.g. one of:
(1) Have a marker comment like "// GEN-POSTMORTEM-METADATA" on the line of the
class definition, or the line before, or maybe reverse its meaning ("//
NO-POSTMORTEM-METADATA") and annotate classes that should be ignored.
(2) Stop making that distinction at all. Would that hurt? Things seem pretty
inconsistent right now anyway.
(3) Have a blacklist (as a Python dictionary) right inside the script, leave
objects.h alone entirely.

[evanlucas, 2015-11-10 11:54:12]

On 2015/11/10 09:38:18, Jakob wrote:
> ACK.
> 
> Semantic meaning of spaces is a horrible, horrible hack. I would appreciate it
> if gen-postmortem-metadata.py could migrate to a saner approach, e.g. one of:
> (1) Have a marker comment like "// GEN-POSTMORTEM-METADATA" on the line of the
> class definition, or the line before, or maybe reverse its meaning ("//
> NO-POSTMORTEM-METADATA") and annotate classes that should be ignored.
> (2) Stop making that distinction at all. Would that hurt? Things seem pretty
> inconsistent right now anyway.
> (3) Have a blacklist (as a Python dictionary) right inside the script, leave
> objects.h alone entirely.

Is that something you would like to see done here instead?

[evanlucas, 2015-11-10 14:31:23]

On 2015/11/10 11:54:12, evan.lucas wrote:
> On 2015/11/10 09:38:18, Jakob wrote:
> > ACK.
> > 
> > Semantic meaning of spaces is a horrible, horrible hack. I would appreciate
it
> > if gen-postmortem-metadata.py could migrate to a saner approach, e.g. one
of:
> > (1) Have a marker comment like "// GEN-POSTMORTEM-METADATA" on the line of
the
> > class definition, or the line before, or maybe reverse its meaning ("//
> > NO-POSTMORTEM-METADATA") and annotate classes that should be ignored.
> > (2) Stop making that distinction at all. Would that hurt? Things seem pretty
> > inconsistent right now anyway.
> > (3) Have a blacklist (as a Python dictionary) right inside the script, leave
> > objects.h alone entirely.
> 
> Is that something you would like to see done here instead?

I submitted https://codereview.chromium.org/1435643002 as an alternative