1) Can I say that is more difficult to infer the source of a wallet owner since address will not be reused after an withdrawal?
Although no one cannot infer the "seed" from addresses, it is still possible to track paths of the transactions. Unless the transaction goes to an exchange address or you use IOTA mixer (see links below), you cannot have a complete anonymity on your transactions.
https://medium.com/iota-ucl/iota-mixer-91f3d39735c1
https://blog.iota.org/research-on-private-transactions-in-iota-cd546751e2c4
2) Is it right to say that is it improbable to attack an address to get to the wallet seed using currently known computational power ?
It is indeed improbable to crack a seed from an address. A seed plus an index (plus a security level) generates a private key. A private key generates an address.
seed + index:0 --> private key:0 --> address:DJAKSDAMSDJA
seed + index:1 --> private key:1 --> address:QPLMGOZYRAPQ
Although forging a private key is possible with available computational power when an user signs for multiple transactions with a private key, in other words, spending more than once from the same address, attacks cannot infer the seed from the private key. The "oneway-ness" of the hashing algorithm makes it impossible to crack an IOTA seed from its private key.