IL level type-remapping, so i8 / i4 are interchangeable, etc. This avoids issues on databases that return 1 as bigint by default, etc

Support of types for Id column in Dapper.Rainbow

From time to time ID column has types different to int. With this change
you could specify column type if it is different to int. Example:
public Table<MyDb, Guid> Users { get; set; }

Minor tweaks to custom type-map implementation; principally, lock-less reading and caching the default type-map

Conflicts:
	Dapper.Rainbow/Database.cs

tweak tests to represent this (and workaround)

Added [ActiveTest] to allow trivial debugging individual tests

Replicated a bug 'Dapper MultiMap doesn't work with splitOn with NULL value'

Fix for issue #70 (support for TimeSpan)

Added dynamic query like for CSharp30.

Also optimised ldloc/lodloca/stloc and fixed enum double-assignment
Signed-off-by: Marc Gravell <marc.gravell@gmail.com>

Conflicts:
	Dapper/SqlMapper.cs

Fix for issue #70
(http://code.google.com/p/dapper-dot-net/issues/detail?id=70) and also
related MS bug with DbType.Time
(https://connect.microsoft.com/VisualStudio/feedback/details/381934/sqlparameter-dbtype-dbtype-time-sets-the-parameter-to-sqldbtype-datetime-instead-of-sqldbtype-time).

Capitalize "id" to maintain consitency

sql_variant support query parameters and SQL Server

Support for ExpandoObject (or at least, "dynamic" and dictionary-like) to be treated directly as a parameter

Updated DynamicParameters to allow inheritance of the AddParameters

Updated DynamicParameters to allow inheritance of the AddParameters method when subclassing for TVPs

Adding some common used overloads for Query, Execute and QueryMultiple

(schema change detection) actually, don't need hash to depend on types; the binding code only considers names, not types - generated code would be the same even if this doesn't change

Rewrite of the schema-change handling; the "see if it errors, then recalc" is, it turns out, stupid.