Rewrite DTM to support dynamic data type configuration

This is basically a rewrite of the DataTypeManager to support dynamic data type configuration.  DTM::Start() has been replaced with DTM::Configure().  Note that the callback is also gone, replaced with notification service style notifications to tell listeners that configuration has started or finished.

I also added some stuff to prepare for adding the "initial download" step for new datatype into the configure process.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=42384

[ncarter (slow), 2010-03-23 20:53:08]

LGTM

http://codereview.chromium.org/1128012/diff/4001/5001
File chrome/browser/sync/glue/data_type_manager.h (right):

http://codereview.chromium.org/1128012/diff/4001/5001#newcode42
chrome/browser/sync/glue/data_type_manager.h:42: typedef
std::set<syncable::ModelType> TypeSet;
Did you consider and reject bitset?

http://codereview.chromium.org/1128012/diff/4001/5003
File chrome/browser/sync/glue/data_type_manager_impl.h (right):

http://codereview.chromium.org/1128012/diff/4001/5003#newcode80
chrome/browser/sync/glue/data_type_manager_impl.h:80:
std::vector<DataTypeController*> needs_stop_;
I like this work queue approach.

http://codereview.chromium.org/1128012/diff/4001/5008
File chrome/browser/sync/profile_sync_service_startup_unittest.cc (right):

http://codereview.chromium.org/1128012/diff/4001/5008#newcode133
chrome/browser/sync/profile_sync_service_startup_unittest.cc:133: class
ProfileSyncServiceStartupBootstrapTest :  // NOLINT
You could put the colon on the next line?
    : public Foo {

[skrul, 2010-03-23 21:23:44]

http://codereview.chromium.org/1128012/diff/4001/5001
File chrome/browser/sync/glue/data_type_manager.h (right):

http://codereview.chromium.org/1128012/diff/4001/5001#newcode42
chrome/browser/sync/glue/data_type_manager.h:42: typedef
std::set<syncable::ModelType> TypeSet;
On 2010/03/23 20:53:08, nick wrote:
> Did you consider and reject bitset?

I considered it but thought that it would be a bit clumsy in practice -- having
to define it as set::bitset<MODEL_TYPE_COUNT> is odd, and IMHO it does not
really convey that it is a group of syncable::ModelType.  Our use of it just
feels more set-ish than bitset-ish :)

http://codereview.chromium.org/1128012/diff/4001/5008
File chrome/browser/sync/profile_sync_service_startup_unittest.cc (right):

http://codereview.chromium.org/1128012/diff/4001/5008#newcode133
chrome/browser/sync/profile_sync_service_startup_unittest.cc:133: class
ProfileSyncServiceStartupBootstrapTest :  // NOLINT
On 2010/03/23 20:53:08, nick wrote:
> You could put the colon on the next line?
>     : public Foo {

Oh yes that does work :)  Thanks :)