"Reinforced diode". Diode connected BJT aka "active diode" is simply a transistor which collector is connected to the base. Thus the collector-emitter part of the transistor is connected in parallel to its base-emitter junction so we can think of this combination as of a "reinforced diode". The current through this "composed diode" is beta times bigger than the current through the single p-n (base-emitter) junction. So its IV curve is more vertical or, as they say, its differential resistance in this part is lower. That is why the active diode is better than the ordinary diode.
Note that the true diode (base-emitter junction) diverts only a beta part of the whole input (collector) current; so it acts as a low power (signal) diode that determines the behavior of the power "diode". Most of the current passes through the collector-emitter junction that initially had the behavior of a current stabilizer but now acts as a voltage stabilizer.
"Reversed" transistor. This connection introduces a voltage-type negative feedback that reverses the transistor behavior. Usually, the input voltage Vbe controls the output collector current Ic of the transistor while here, thanks to the negative feedback, it seems as if the "input" collector current controls the "output" voltage Vbe. This "reversed" transistor is used in the input part of the simple BJT current mirror (QREF in the Bimpelrekkie's picture).
This "reversal trick" can be observed in any negative feedback system since it adjusts its input so that to obtain the desired output. As a result, the output becomes an input and the input becomes an output. Another typical example is the ubiquitous op-amp non-inverting amplifier where the op-amp adjusts the input voltage VOA of the R1-R2 voltage divider so that to make its output voltage VR1 = VOA.R1/(R1 + R2) equal to the true input voltage VIN. As a result, the attenuator acts (with the help of the op-amp) as an amplifier with a gain of (R1 + R2)/R1.
"Rubber Zener diode". If we apply not the whole collector-emitter voltage to the base-emitter junction but a part of it, VBE will be multiplied (like in the non-inverting amplifier). The "transistor diode" will act as a "transistor Zener diode" with any desired voltage. This network is widely used as a bias circuit in op-amp and power amplifiers.
Could you please throw some more light on "voltage-type negative feedback"?
The transistor and the collector resistor form the classic common-emitter amplifying stage. This is a voltage amplifier where we apply the input voltage to its input port - the base-emitter junction, and take the output voltage from its output port - the collector-emitter junction. Since the ground is common, when we connect the collector to base, actually we connect the output port to the input port in parallel... simply, the output to the input... As a result, all the output (collector) voltage is applied to the input; hence the name "voltage-type". Applied in such a "parallel" (shunt) way, the output voltage make the transistor decrease the same output voltage until the equilibrium is reached (roughly, VC = VB = 0.65 V). The name of this mechanism is "negative feedback"... and here it is a "voltage-type negative feedback".
Experiments
Ordinary diode
simulate this circuit – Schematic created using CircuitLab
Base-emitter junction
simulate this circuit
"Active diode"
simulate this circuit
Zener diode
simulate this circuit
"Rubber Zener diode"
simulate this circuit
Constant-current 'diode'
simulate this circuit