Smothered Mate: Inspired by <Eggman>'s comment, here's analysis from a _DTC_ tablebase (http://chess.jaet.org/endings/), rather than a DTM tablebase:
1 move = 2 ply
PTC = ply to conversion , _not_ counting
the capturing or checkmating move
So, just after a capture,
PTC = 99 would be a win and PTC = 100 would be a draw.
71. ... Kxb5 entered the KQKR ending.
PTC = 44
72. Qc8 slows down conversion by 1 move.
Kf6 and Ke6 were the PTC-optimal moves.
PTC = 45
72. ... Rd5 speeds up conversion by 5 moves.
Kb4 was uniquely PTC-optimal.
PTC = 34
73. Ke6 slows down conversion by 4 moves.
Qc3 was the only other move that does not slow it even more.
PTC = 41
73. ... Rd4 speeds up conversion by 2 moves.
Rc5 was uniquely PTC-optimal.
PTC = 36
74. Ke5 Rd3 was PTC-optimal, but not uniquely so.
PTC = 34
75. Qc2 Rd8 76. Qb3+ Kc5 77. Qc3+ was uniquely PTC-optimal.
PTC = 29
77. ... Kb5 speeds up conversion by 2 moves.
(There's only one other legal move.)
PTC = 24
78. Ke6 Kb6 79. Qc4 Rg8 was uniquely PTC-optimal.
PTC = 20
80. Qd4+ was PTC-optimal.
Qb4+ was the only other PTC-optimal move.
PTC = 19
80. ... Kc6 was uniquely PTC-optimal.
PTC = 18
81. Qc3+ slows down conversion by 2 moves.
The only moves to convert faster were Qe4+,Qf6,Ke7.
Those 3 moves all give the same PTC.
PTC = 21
81. ... Kb5 was uniquely PTC-optimal.
PTC = 20
82. Kd6 slows down conversion by 10 moves.
Qe5+ was uniquely PTC-optimal.
PTC = 39
82. ... Rg6+ was uniquely PTC-optimal.
Rg4 was the unique _next_-best, and
would've sped up conversion by 9 moves.
PTC = 38
83. Kc7 was PTC-optimal.
Kd7 was the only other PTC-optimal move.
PTC = 37
83. ... Rg4 was uniquely PTC-optimal.
Ra6 was unique _next_-best, and
would've sped up conversion by 13 moves.
PTC = 36
84. Qc6+ slows down conversion by 2 moves.
Qe3 was uniquely PTC-optimal.
PTC = 39
84. ... Kb4 was uniquely PTC-optimal.
PTC = 38
85. Qd6+ slows down conversion by 3 moves.
Qb6+ was uniquely PTC-optimal.
PTC = 43
85. ... Kc3 86. Kc6 was uniquely PTC-optimal.
Ka5 was black's unique _next_-best, and only lasts 4 ply before the rook is taken.
PTC = 41
86. ... Rd4 was PTC-optimal.
Rc4+ was the only other PTC-optimal move.
PTC = 40
87. Qa3+ slows down conversion by 1 move.
Qg3+ was the only move that converts faster.
PTC = 41
87. ... Kd2 was uniquely PTC-optimal.
PTC = 40
88. Kc5 slows down conversion by 1 move, _and_ reaches
a rotation of the position from just after 73. Ke6 .
Qf3 was the only other move that does not slow it even more.
88. ... Re4 89. Kd5 are the corresponding rotations
of the corresponding moves from that position.
89. ... Rg4 speeds up conversion by 1 move.
Rf4 was PTC-optimal (but not uniquely so)
_and_ would've kept the rotation.
PTC = 32
90. Qf3 Rb4 91. Kc5 Ra4 was uniquely PTC-optimal.
PTC = 28
92. Qf6 slows down conversion by 6 moves.
Qf2+ was uniquely PTC-optimal.
PTC = 39
92. ... Kd3 speeds up conversion by 6 moves.
Ke3 was uniquely PTC-optimal.
PTC = 26
93. Qd6+ slows down conversion by 9 moves.
Qf1+ was uniquely PTC-optimal.
PTC = 43
93. ... Ke3 was uniquely PTC-optimal.
Kc3 was unique _next_-best, and
would've sped up conversion by 8 moves.
PTC = 42
94. Qg3+ was PTC-optimal.
Qd1 and Qh6+ were the other PTC-optimal moves.
PTC = 41
94. ... Ke2 was uniquely PTC-optimal.
PTC = 40
95. Qc3 Rf4 were PTC-optimal, but not uniquely so.
PTC = 38
96. Kd5 was uniquely PTC-optimal.
PTC = 37
96. ... Rg4 speeds up conversion by 4 moves.
Rh4 was uniquely PTC-optimal.
PTC = 28
97. Ke5 was uniquely PTC-optimal.
PTC = 27