Bonding is so multi faults will not cause two exposed metallic objects from have a voltage between them. If the supply on feeds one item for example a shaver then it would be safer not to have bonding.
That statement might be confusing to those who are not 'in the know', since I presume you are talking specifically about an 'isolating transformer', the output of which is 'floating' (neither side connected to earth) - which, in UK installations, is usually only seen in 'shaver sockets'. Hence ....
But most transformers supply many things, so we use bonding to cause auto disconnection in the case of a fault.
Yes, but the DNO transformers which supply installations, hence supply 'many things',
DO have one side connected to earth, so it makes no difference whether they are supplying one or many things.
So a MCB is a duel device, both thermal and magnetic, the latter is much faster, but is set 5 x (B), 10 x (C) or 20 x (D) to the thermal device, so an immersion heater (3 kW) is likely supplied with a 16B MCB which will need 16 x 5 = 90 amps to trip, so 230/90 = 2.56 Ω with 5% for safety that's 2.44 Ω if using earth rods that would be hard to achieve to earth, so only option is a RCD. But still working with 2.44 Ω live to live or line to neutral.
Again, I think this might all confuse people. For a start, your arithmetic, and basis of your calculations, could do with some improvement. Firstly, 16 x 5 is 80, not 90. Secondly, in calculating the maximum Zs which would be guaranteed to result in magnetic tripping (in the face of a zero impedance L-CPC fault) you should take Cmin (0.95, even though it should really be 0.94

) into account and calculate that figure using a voltage of 218.5 V (230 x 0.95), not 230 V. The correct calculations would therefore have led to a minimum current for guaranteed magnetic tripping (In x 5) of 80 A, hence a maximum Zs of 2.73 Ω (218.5 / 80).
Thirdly, and perhaps most importantly, although what you say about 'using earth rods' is correct, it would probably be desrable to point out that this is true of
any circuit in a TT installation - so, adequate fault protection of every/any cuircuit in a TT installation would have to be provided by an RCD.
....But still working with 2.44 Ω live to live or line to neutral.
Even I am confused by that one, so good luck to those who know less than I do

What do you mean by "live to live" - do you mean "phase to phase" in a 3-phase installation, "Line/Phase to Neutral" or what? .. and what does the sentence as a whole mean?
In the days of fuses, if we got is slightly out, it was not so bad, it took a few mS longer to disconnect, but with a MCB the difference between magnetic disconnect time and thermal disconnect time is huge ....
I don't think that even "a few" (or even "a good few"!) seconds would make much difference (it's incredibly improbable that someone would be 'touching the wrong things' at the very moment an L-E fault arose), and currents approaching the magnetic trip threshold will thermally trip an MCB in well under 15 seconds (these curves being for B-curve MCBs) ...