If you remove the top five bonuses and plot the remaining bonuses vs. slot recommendations, the results are surprisingly consistent:   bonus = 1.25 * recommendation, with an r^2 > 0.96.  I wonder if previous drafts adhered to a similar relationship – scrap that top few crazy, skewing bonuses, and the remaining bunch has a strong linear relationship to the slot recommendations.

why is #2 pick Jameson Taillon's slot bonus $750K more than #1 pick Bryce Harper's?

It appears as if people don't get it.  He is not comparing individual player bonuses to their respective slots.  He is comparing the top 50 actual bonuses received to the top 50 slot recommendations.

Now I get it. 
As to thet question about Tallion and Harper.  Harper has a ML contract, so he gets more $ as a wage over the contract. Tallion signed the normal rookie contract, to make up for the disparity in wage, he was given more upfront. 

If you take the economic view of this, just because teams are paying over slot does not mean that the prospects are worth that.   It could be an imbalance in the economic system.  Markets are not always efficient, they are simply efficient over time
In any case, this is not a true market anyway, in the economic sense.  The draft is an artificially imbalanced market to favor the worse performing teams.  And with only one draft per year, it could take a long time for equilibrium, if such can exist in such a market, to be reached.
The players have long had leverage that most agents do not take advantage of, to squeeze more money out of baseball.  Boras was just the first to realize that and to capitalize on that inefficiency in this market.  Others have followed, and basically follow the quote above as their mantra.  
However, I would argue that is a falsehood for all but the very top picks in the first round.  I recall one Giants pick (Wendell Fairley) who waited until the last day to sign, and he barely got maybe $25-50K more than slot (I think he wanted $1M as an even significant amount), but his pro career might be going better now had he signed early for slot and gotten more playing time his first season.
This artificial draft  was created to avoid the top teams from simply buying all the best prospects (read, Yankees).  Of course, some could argue that the top teams are doing it now, but the odds of prospects becoming a good player in the majors is so low the teams would have to buy a lot to guarantee any immediate return, and in any case the draft still does a good job of limiting that and in keeping the best prospects out of their hands, at least until free agency.
And people misjudge what a bonus really represent in the costs of finding and developing a good player in the majors.  It is not $2M or $4M or $6M for a player, there is a probability for each player that he might become a good player, not all do just because they are paid $2M.  
My calculations of the chances of finding a good player in the draft places the true acquisition costs of a good player at somewhere in the $10-15M range (meaning that you pay bonuses of around $10-15M to find each good player, on average).  When the average team spends around $5M per year, that works out to finding a good player, on average, every 2-3 years.  
And losing teams have a huge advantage, the odds of finding a good player is double and even quadruple that of a pick in the last third of the draft, which is where all the team contending for the playoffs tend to end up in the first round.  Roughly 10% chance for that back of 1st round pick, roughly 20% for picks 6-10, roughly 40% for the 1-5 overall.   But that is why the bonuses are that much higher there.
That's probably why the bonuses tend to spike early in the draft, because slotting is a linear function while the odds of finding a good player is exponential.  Things are roughly linear past the first round because the odds are so low anyhow. 

Chris Sale is a unique situation.  Only one like that in this year's draft.