clean up names in zipstater

We were using some sort of special malformed non-path seperator character to stand in for the path seperator character in in the lump names, and then compensating for that brokenness when loading lumps.
On save, we were relying in our zip library compensating for the brokenness.

src/BizHawk.Client.Common/savestates/BinaryStateLump.cs

@@ -51,19 +51,19 @@ namespace BizHawk.Client.Common
 		public static BinaryStateLump Session { get; private set; }
 
 		// branch stuff
-		[Name("Branches\\CoreData", "bin")]
+		[Name("Branches/CoreData", "bin")]
 		public static BinaryStateLump BranchCoreData { get; private set; }
-		[Name("Branches\\InputLog", "txt")]
+		[Name("Branches/InputLog", "txt")]
 		public static BinaryStateLump BranchInputLog { get; private set; }
-		[Name("Branches\\FrameBuffer", "bmp")]
+		[Name("Branches/FrameBuffer", "bmp")]
 		public static BinaryStateLump BranchFrameBuffer { get; private set; }
-		[Name("Branches\\CoreFrameBuffer", "bmp")]
+		[Name("Branches/CoreFrameBuffer", "bmp")]
 		public static BinaryStateLump BranchCoreFrameBuffer { get; private set; }
-		[Name("Branches\\Header", "json")]
+		[Name("Branches/Header", "json")]
 		public static BinaryStateLump BranchHeader { get; private set; }
-		[Name("Branches\\Markers", "txt")]
+		[Name("Branches/Markers", "txt")]
 		public static BinaryStateLump BranchMarkers { get; private set; }
-		[Name("Branches\\UserText", "txt")]
+		[Name("Branches/UserText", "txt")]
 		public static BinaryStateLump BranchUserText { get; private set; }
 
 		[AttributeUsage(AttributeTargets.Property)]

src/BizHawk.Client.Common/savestates/ZipStateLoader.cs

@@ -62,7 +62,7 @@ namespace BizHawk.Client.Common
 					name = name.Substring(0, i);
 				}
 
-				_entriesByName.Add(name.Replace('/', '\\'), z);
+				_entriesByName.Add(name, z);
 			}
 		}