In my case the concept of mean reversion is driven by the fact, that if a market or asset class has outperformed, then over time it will tend to revert. This is more applicable to the index than an a stock & as Taleb pointed out purely driven by probability scenarios.
For stocks there may be a case of serious rerating because the business qualities are discovered and the market identifies it as an above average business with a good management eg Ajanta Pharma, Eicher Motors. Now as you suggested that if a stock like Lupin has grown at 30% over the years, it could also be that the earning may also have compounded 25-30% over the years. Then there is no reason to be alarmed and sell. However over a period of time if a stock you bought at 1 times P/B is now say 10 times P/B, eventhough the PE has increased from say 5 to 25 now, I would tend to be cautious.
The rule I follow of a stock not being allowed to be more than 25% of portfolio is driven by my past experience, where one stock which had hit more than 50% of my portfolio crashed 92% from the peak in 2008 Lehman Crises. Its an event I never want to experience again and hence the rule
So to answer you question the mean reversion analysis is more driven by the overall values in the market, rather then the sector or a stock. Also I tend to take money of the table when times are good and tend to invest in the safer categories like Debt, bonds.