Why was the Planck constant $h$ fixed to be exactly $6.62607015\times10^{−34}\text{Js}$ and not some other value?
There are many, many instruments which are calibrated using the old definition of the kilogramme - the International Prototype Kilogramme (IPK) made of a platinum-iridium alloy.
So one kilogramme measured using the new definition had to be as close as possible to one kilogramme using old definition so as not to have to recalibrate all instruments which relied on the old definition of the kilogramme.
Using the old definition of the kilogramme (IPK) the numerical value of Planck’s constant was measured as accurately as possible using the Kibble (watt) balance and the X-ray crystal density method.
The two values that you quoted $6.62606957 \times 10^{−34}\, \rm kg\,m^2s^{-1}$ and later $ 6.62607015 \times 10^{−34}\, \rm kg\,m^2s^{-1}$ were the results of such measurements of Planck's constant.
As of 20 May 2019 the determination/definition was turned on its head with the value of Planck’s constant defined as $ 6.62607015 \times 10^{−34}\, \rm kg\,m^2s^{-1}$ and the IPK (made of a platinum-iridium alloy) having a measured value of one kilogramme to within one part in $10$ billion.
On page 131 of the BIPM brochure on the SI system of units it states:
The number chosen for the numerical value of the Planck constant in this definition is such that at the time of its adoption, the kilogram was equal to the mass of the international prototype, m(K) = 1 kg, with a relative standard uncertainty of $1 \times 10^{-8}$, which was the standard uncertainty of the combined best estimates of the value of the Planck constant at that time.
In future the new definition of one kilogramme via the defined exact value of Planck's constant will enable measurements to be made to see by how much the masses of the IPK and its daughters change with time.
—-
why did they choose this arbitrary value of 6.62607015×10−34.
This value was chosen to make one kilogramme using the old definition (IPK) as close as possible to one kilogramme using the new definition (via Planck's constant).
instead of something more exact like 6.62607x10-34.
This would have required the recalibration of many, many (accurate) instruments.
It is an updated experimental value that matches measurements by Kibble balances and by counting atoms in silicon spheres to determine Avogadro’s number. The 2010 measurement was presumably consistent with only one metrological approach. The new exact value is consistent with both.
See this NIST page, which says
The Kibble balance and Avogadro project measurements are not so much competing with as complementing each other to define the kilogram. Measurements from both experiments were used to determine the final value of $h$ for the redefined SI. That final value is $6.626070150\times 10^{-34}\text{ kg}\cdot\text{m}^2/\text{s}$.