Many thanks Micawber for this interesting analyze. However, Mucklow was the first to try 6.e5 even if we find the game McDonnell-De la Bourdonnais, London, 1834 by transpose again.

What do you think of following move after 3.Bc4 Qh4+ 4.Kf1 g5 5.Nc3 Bg7 ?

4) 6. Nf3 Qh5  (6... Qh6 perhaps) 7.h4!? h6 8.Nd5 (8.d4 e6 9.e5 += Anderssen-Slous, London, 1851) Kd8 9.d4

No I havent met this variation..

3.Bc4,Qh4+ 4.Kf1,g5 5.Nc3,Bg7 6.e5

Three observations:

(1) I dont regard 3....Qh4+ as the best answer for black against the
KGA-bishop gambit. IMHO better options are 
3.....Nc6 or 3....Nf6 4.Nc3,c6 are more solid options.

(2) If black prefers 3...Qh4+ he may consider other variations:
4.Kf1,d6 5.d4, Be6!? 6.Qd3, Nd7! seems playable.

(3) There are some examples of your variation from a later date,
and a transposition from an earlier date:
4.Kf1,g5 5.Nc3,Bg7 6.d4! (iso 6.e5)

(3a) 6....d6 7.e5   (Andersen-Blackburn, (London 1862)
note that this position might be reached from 6.e5,d6 7.d4 as well

(3b)   
6.......Nge7
7.g3 
(a standard method to break up the kingside)
7....., fxg3
8.Kg2, Qh6
(Keres advised 8....d6 when white is a little better)
9.Nf3 , d5
10.hxg3,Qg6
Reti-Spielmann, Trebitsch 1910.
White has the better chances, 
Spielmann probably felt that he was on the wrong side of the board 

(3c)
6.......Nc6!?
(interestingly not mentioned in Johansson's monumental work)
7.e5, Nge7!
8.Nf3,Qh5
9.Ne4,h6
(now we have transferred into a position from the "Mucklow game"
10.Nf6, Bxf6
11.exf6,d5!
(better than Ranken's  11...Nf5 as Breitazao allready indicated)
12.Bd3, Nf5  with and advantage for black

MacDonnel-De la Bourdonnais, London, 1834
Predating the Mucklow game by 20 years!
So if this gambit should be named
MacDonnel's gambit is a more appropiate name.
Looking at the game made me wonder if the gambit
6.d4,Nc6 7.g3!? is  viable

Finally after 6.d4,Nc6
7.Nf3,Qh5 8.Nd5! might be an improvement as
8....Kd8 9.Be2 seems good for white:
(threats are Bh5 and Nxg5 followed by Bxf4 and Bxc7+)